diff --git a/clang/include/clang/Basic/BuiltinsAMDGPU.def b/clang/include/clang/Basic/BuiltinsAMDGPU.def index ab432b7a8ad58..f9e4785eacfc3 100644 --- a/clang/include/clang/Basic/BuiltinsAMDGPU.def +++ b/clang/include/clang/Basic/BuiltinsAMDGPU.def @@ -350,6 +350,31 @@ BUILTIN(__builtin_amdgcn_endpgm, "v", "nr") BUILTIN(__builtin_amdgcn_get_fpenv, "WUi", "n") BUILTIN(__builtin_amdgcn_set_fpenv, "vWUi", "n") +//===----------------------------------------------------------------------===// + +// Wave Reduction builtins. + +//===----------------------------------------------------------------------===// + +BUILTIN(__builtin_amdgcn_wave_reduce_add_i32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_sub_i32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_min_i32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_min_u32, "UiUii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_max_i32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_max_u32, "UiUii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_and_b32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_or_b32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_xor_b32, "iii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_add_i64, "WiWii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_sub_i64, "WiWii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_min_i64, "WiWii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_min_u64, "WUiWUii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_max_i64, "WiWii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_max_u64, "WUiWUii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_and_b64, "WiWii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_or_b64, "WiWii", "nc") +BUILTIN(__builtin_amdgcn_wave_reduce_xor_b64, "WiWii", "nc") + //===----------------------------------------------------------------------===// // R600-NI only builtins. //===----------------------------------------------------------------------===// diff --git a/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp b/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp index bd44874eac470..a4386b3399353 100644 --- a/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp +++ b/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp @@ -295,11 +295,69 @@ void CodeGenFunction::AddAMDGPUFenceAddressSpaceMMRA(llvm::Instruction *Inst, Inst->setMetadata(LLVMContext::MD_mmra, MMRAMetadata::getMD(Ctx, MMRAs)); } +static Intrinsic::ID getIntrinsicIDforWaveReduction(unsigned BuiltinID) { + switch (BuiltinID) { + default: + llvm_unreachable("Unknown BuiltinID for wave reduction"); + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_add_i32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_add_i64: + return Intrinsic::amdgcn_wave_reduce_add; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_sub_i32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_sub_i64: + return Intrinsic::amdgcn_wave_reduce_sub; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_min_i32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_min_i64: + return Intrinsic::amdgcn_wave_reduce_min; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_min_u32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_min_u64: + return Intrinsic::amdgcn_wave_reduce_umin; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_max_i32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_max_i64: + return Intrinsic::amdgcn_wave_reduce_max; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_max_u32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_max_u64: + return Intrinsic::amdgcn_wave_reduce_umax; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_and_b32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_and_b64: + return Intrinsic::amdgcn_wave_reduce_and; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_or_b32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_or_b64: + return Intrinsic::amdgcn_wave_reduce_or; + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_xor_b32: + case clang::AMDGPU::BI__builtin_amdgcn_wave_reduce_xor_b64: + return Intrinsic::amdgcn_wave_reduce_xor; + } +} + Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID, const CallExpr *E) { llvm::AtomicOrdering AO = llvm::AtomicOrdering::SequentiallyConsistent; llvm::SyncScope::ID SSID; switch (BuiltinID) { + case AMDGPU::BI__builtin_amdgcn_wave_reduce_add_i32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_sub_i32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_min_i32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_min_u32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_max_i32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_max_u32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_and_b32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_or_b32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_xor_b32: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_add_i64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_sub_i64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_min_i64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_min_u64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_max_i64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_max_u64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_and_b64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_or_b64: + case AMDGPU::BI__builtin_amdgcn_wave_reduce_xor_b64: { + Intrinsic::ID IID = getIntrinsicIDforWaveReduction(BuiltinID); + llvm::Value *Value = EmitScalarExpr(E->getArg(0)); + llvm::Value *Strategy = EmitScalarExpr(E->getArg(1)); + llvm::Function *F = CGM.getIntrinsic(IID, {Value->getType()}); + return Builder.CreateCall(F, {Value, Strategy}); + } case AMDGPU::BI__builtin_amdgcn_div_scale: case AMDGPU::BI__builtin_amdgcn_div_scalef: { // Translate from the intrinsics's struct return to the builtin's out diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn.cl index bf022bc6eb446..16f5a524f3094 100644 --- a/clang/test/CodeGenOpenCL/builtins-amdgcn.cl +++ b/clang/test/CodeGenOpenCL/builtins-amdgcn.cl @@ -398,6 +398,384 @@ void test_s_sendmsghalt_var(int in) __builtin_amdgcn_s_sendmsghalt(1, in); } +// CHECK-LABEL: @test_wave_reduce_add_i32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.add.i32( +void test_wave_reduce_add_i32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_add_i32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_add_i64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.add.i64( +void test_wave_reduce_add_i64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_add_i64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_add_i32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.add.i32( +void test_wave_reduce_add_i32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_add_i32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_add_i64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.add.i64( +void test_wave_reduce_add_i64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_add_i64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_add_i32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.add.i32( +void test_wave_reduce_add_i32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_add_i32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_add_i64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.add.i64( +void test_wave_reduce_add_i64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_add_i64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_sub_i32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.sub.i32( +void test_wave_reduce_sub_i32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_sub_i32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_sub_i64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.sub.i64( +void test_wave_reduce_sub_i64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_sub_i64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_sub_i32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.sub.i32( +void test_wave_reduce_sub_i32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_sub_i32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_sub_i64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.sub.i64( +void test_wave_reduce_sub_i64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_sub_i64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_sub_i32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.sub.i32( +void test_wave_reduce_sub_i32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_sub_i32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_sub_i64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.sub.i64( +void test_wave_reduce_sub_i64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_sub_i64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_and_b32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.and.i32( +void test_wave_reduce_and_b32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_and_b32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_and_b64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.and.i64( +void test_wave_reduce_and_b64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_and_b64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_and_b32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.and.i32( +void test_wave_reduce_and_b32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_and_b32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_and_b64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.and.i64( +void test_wave_reduce_and_b64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_and_b64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_and_b32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.and.i32( +void test_wave_reduce_and_b32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_and_b32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_and_b64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.and.i64( +void test_wave_reduce_and_b64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_and_b64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_or_b32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.or.i32( +void test_wave_reduce_or_b32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_or_b32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_or_b64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.or.i64( +void test_wave_reduce_or_b64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_or_b64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_or_b32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.or.i32( +void test_wave_reduce_or_b32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_or_b32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_or_b64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.or.i64( +void test_wave_reduce_or_b64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_or_b64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_or_b32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.or.i32( +void test_wave_reduce_or_b32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_or_b32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_or_b64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.or.i64( +void test_wave_reduce_or_b64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_or_b64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_xor_b32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.xor.i32( +void test_wave_reduce_xor_b32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_xor_b32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_xor_b64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.xor.i64( +void test_wave_reduce_xor_b64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_xor_b64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_xor_b32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.xor.i32( +void test_wave_reduce_xor_b32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_xor_b32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_xor_b64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.xor.i64( +void test_wave_reduce_xor_b64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_xor_b64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_xor_b32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.xor.i32( +void test_wave_reduce_xor_b32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_xor_b32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_xor_b64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.xor.i64( +void test_wave_reduce_xor_b64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_xor_b64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_min_i32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.min.i32( +void test_wave_reduce_min_i32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_min_i32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_min_i64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.min.i64( +void test_wave_reduce_min_i64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_min_i64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_min_i32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.min.i32( +void test_wave_reduce_min_i32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_min_i32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_min_i64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.min.i64( +void test_wave_reduce_min_i64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_min_i64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_min_i32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.min.i32( +void test_wave_reduce_min_i32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_min_i32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_min_i64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.min.i64( +void test_wave_reduce_min_i64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_min_i64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_min_u32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.umin.i32( +void test_wave_reduce_min_u32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_min_u32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_min_u64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.umin.i64( +void test_wave_reduce_min_u64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_min_u64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_min_u32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.umin.i32( +void test_wave_reduce_min_u32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_min_u32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_min_u64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.umin.i64( +void test_wave_reduce_min_u64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_min_u64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_min_u32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.umin.i32( +void test_wave_reduce_min_u32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_min_u32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_min_u64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.umin.i64( +void test_wave_reduce_min_u64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_min_u64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_max_i32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.max.i32( +void test_wave_reduce_max_i32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_max_i32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_max_i64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.max.i64( +void test_wave_reduce_max_i64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_max_i64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_max_i32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.max.i32( +void test_wave_reduce_max_i32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_max_i32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_max_i64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.max.i64( +void test_wave_reduce_max_i64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_max_i64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_max_i32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.max.i32( +void test_wave_reduce_max_i32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_max_i32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_max_i64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.max.i64( +void test_wave_reduce_max_i64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_max_i64(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_max_u32_default +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.umax.i32( +void test_wave_reduce_max_u32_default(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_max_u32(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_max_u64_default +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.umax.i64( +void test_wave_reduce_max_u64_default(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_max_u64(in, 0); +} + +// CHECK-LABEL: @test_wave_reduce_max_u32_iterative +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.umax.i32( +void test_wave_reduce_max_u32_iterative(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_max_u32(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_max_u64_iterative +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.umax.i64( +void test_wave_reduce_max_u64_iterative(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_max_u64(in, 1); +} + +// CHECK-LABEL: @test_wave_reduce_max_u32_dpp +// CHECK: {{.*}}call{{.*}} i32 @llvm.amdgcn.wave.reduce.umax.i32( +void test_wave_reduce_max_u32_dpp(global int* out, int in) +{ + *out = __builtin_amdgcn_wave_reduce_max_u32(in, 2); +} + +// CHECK-LABEL: @test_wave_reduce_max_u64_dpp +// CHECK: {{.*}}call{{.*}} i64 @llvm.amdgcn.wave.reduce.umax.i64( +void test_wave_reduce_max_u64_dpp(global int* out, long in) +{ + *out = __builtin_amdgcn_wave_reduce_max_u64(in, 2); +} + // CHECK-LABEL: @test_s_barrier // CHECK: {{.*}}call{{.*}} void @llvm.amdgcn.s.barrier( void test_s_barrier() diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp index 6cf2055c8e565..975264ab55b0d 100644 --- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp +++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp @@ -5069,19 +5069,28 @@ static MachineBasicBlock *emitIndirectDst(MachineInstr &MI, static uint32_t getIdentityValueForWaveReduction(unsigned Opc) { switch (Opc) { case AMDGPU::S_MIN_U32: + case AMDGPU::V_CMP_LT_U64_e64: // umin.u64 return std::numeric_limits::max(); case AMDGPU::S_MIN_I32: + case AMDGPU::V_CMP_LT_I64_e64: // min.i64 return std::numeric_limits::max(); case AMDGPU::S_MAX_U32: + case AMDGPU::V_CMP_GT_U64_e64: // umax.u64 return std::numeric_limits::min(); case AMDGPU::S_MAX_I32: + case AMDGPU::V_CMP_GT_I64_e64: // max.i64 return std::numeric_limits::min(); case AMDGPU::S_ADD_I32: + case AMDGPU::S_ADD_U64_PSEUDO: case AMDGPU::S_SUB_I32: + case AMDGPU::S_SUB_U64_PSEUDO: case AMDGPU::S_OR_B32: + case AMDGPU::S_OR_B64: case AMDGPU::S_XOR_B32: + case AMDGPU::S_XOR_B64: return std::numeric_limits::min(); case AMDGPU::S_AND_B32: + case AMDGPU::S_AND_B64: return std::numeric_limits::max(); default: llvm_unreachable("Unexpected opcode in getIdentityValueForWaveReduction"); @@ -5102,26 +5111,38 @@ static MachineBasicBlock *lowerWaveReduce(MachineInstr &MI, bool isSGPR = TRI->isSGPRClass(MRI.getRegClass(SrcReg)); Register DstReg = MI.getOperand(0).getReg(); MachineBasicBlock *RetBB = nullptr; + bool is32BitOpc = TRI->getRegSizeInBits(*MRI.getRegClass(DstReg)) == 32; if (isSGPR) { switch (Opc) { case AMDGPU::S_MIN_U32: + case AMDGPU::V_CMP_LT_U64_e64: /*umin*/ case AMDGPU::S_MIN_I32: + case AMDGPU::V_CMP_LT_I64_e64: /*min*/ case AMDGPU::S_MAX_U32: + case AMDGPU::V_CMP_GT_U64_e64: /*umax*/ case AMDGPU::S_MAX_I32: + case AMDGPU::V_CMP_GT_I64_e64: /*max*/ case AMDGPU::S_AND_B32: - case AMDGPU::S_OR_B32: { + case AMDGPU::S_AND_B64: + case AMDGPU::S_OR_B32: + case AMDGPU::S_OR_B64: { // Idempotent operations. - BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MOV_B32), DstReg).addReg(SrcReg); + unsigned movOpc = is32BitOpc ? AMDGPU::S_MOV_B32 : AMDGPU::S_MOV_B64; + BuildMI(BB, MI, DL, TII->get(movOpc), DstReg).addReg(SrcReg); RetBB = &BB; break; } case AMDGPU::S_XOR_B32: + case AMDGPU::S_XOR_B64: case AMDGPU::S_ADD_I32: - case AMDGPU::S_SUB_I32: { + case AMDGPU::S_ADD_U64_PSEUDO: + case AMDGPU::S_SUB_I32: + case AMDGPU::S_SUB_U64_PSEUDO: { const TargetRegisterClass *WaveMaskRegClass = TRI->getWaveMaskRegClass(); const TargetRegisterClass *DstRegClass = MRI.getRegClass(DstReg); Register ExecMask = MRI.createVirtualRegister(WaveMaskRegClass); - Register ActiveLanes = MRI.createVirtualRegister(DstRegClass); + Register ActiveLanes = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); bool IsWave32 = ST.isWave32(); unsigned MovOpc = IsWave32 ? AMDGPU::S_MOV_B32 : AMDGPU::S_MOV_B64; @@ -5136,21 +5157,68 @@ static MachineBasicBlock *lowerWaveReduce(MachineInstr &MI, .addReg(Exec->getOperand(0).getReg()); switch (Opc) { - case AMDGPU::S_XOR_B32: { + case AMDGPU::S_XOR_B32: + case AMDGPU::S_XOR_B64: { // Performing an XOR operation on a uniform value // depends on the parity of the number of active lanes. // For even parity, the result will be 0, for odd // parity the result will be the same as the input value. - Register ParityRegister = MRI.createVirtualRegister(DstRegClass); + Register ParityRegister = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); auto ParityReg = BuildMI(BB, MI, DL, TII->get(AMDGPU::S_AND_B32), ParityRegister) .addReg(NewAccumulator->getOperand(0).getReg()) - .addImm(1); - BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), DstReg) - .addReg(SrcReg) - .addReg(ParityReg->getOperand(0).getReg()); - break; + .addImm(1) + .setOperandDead(3); // Dead scc + if (is32BitOpc) { + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), DstReg) + .addReg(SrcReg) + .addReg(ParityReg->getOperand(0).getReg()); + break; + } else { + Register DestSub0 = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register DestSub1 = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register Op1H_Op0L_Reg = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register CarryReg = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + + const TargetRegisterClass *SrcRC = MRI.getRegClass(SrcReg); + const TargetRegisterClass *SrcSubRC = + TRI->getSubRegisterClass(SrcRC, AMDGPU::sub0); + + MachineOperand Op1L = TII->buildExtractSubRegOrImm( + MI, MRI, MI.getOperand(1), SrcRC, AMDGPU::sub0, SrcSubRC); + MachineOperand Op1H = TII->buildExtractSubRegOrImm( + MI, MRI, MI.getOperand(1), SrcRC, AMDGPU::sub1, SrcSubRC); + + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), DestSub0) + .add(Op1L) + .addReg(ParityReg->getOperand(0).getReg()); + + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), Op1H_Op0L_Reg) + .add(Op1H) + .addReg(ParityReg->getOperand(0).getReg()); + + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_HI_U32), CarryReg) + .add(Op1L) + .addReg(ParityReg->getOperand(0).getReg()); + + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_ADD_U32), DestSub1) + .addReg(CarryReg) + .addReg(Op1H_Op0L_Reg) + .setOperandDead(3); // Dead scc + + BuildMI(BB, MI, DL, TII->get(TargetOpcode::REG_SEQUENCE), DstReg) + .addReg(DestSub0) + .addImm(AMDGPU::sub0) + .addReg(DestSub1) + .addImm(AMDGPU::sub1); + break; + } } case AMDGPU::S_SUB_I32: { Register NegatedVal = MRI.createVirtualRegister(DstRegClass); @@ -5171,6 +5239,76 @@ static MachineBasicBlock *lowerWaveReduce(MachineInstr &MI, .addReg(NewAccumulator->getOperand(0).getReg()); break; } + case AMDGPU::S_ADD_U64_PSEUDO: + case AMDGPU::S_SUB_U64_PSEUDO: { + Register DestSub0 = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register DestSub1 = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register Op1H_Op0L_Reg = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register Op1L_Op0H_Reg = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register CarryReg = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register AddReg = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register NegatedValLo = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register NegatedValHi = + MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + + const TargetRegisterClass *Src1RC = MRI.getRegClass(SrcReg); + const TargetRegisterClass *Src1SubRC = + TRI->getSubRegisterClass(Src1RC, AMDGPU::sub0); + + MachineOperand Op1L = TII->buildExtractSubRegOrImm( + MI, MRI, MI.getOperand(1), Src1RC, AMDGPU::sub0, Src1SubRC); + MachineOperand Op1H = TII->buildExtractSubRegOrImm( + MI, MRI, MI.getOperand(1), Src1RC, AMDGPU::sub1, Src1SubRC); + + if (Opc == AMDGPU::S_SUB_U64_PSEUDO) { + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), NegatedValLo) + .addReg(NewAccumulator->getOperand(0).getReg()) + .addImm(-1); + + MachineInstr *NegatedHi = + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_ASHR_I32), NegatedValHi) + .addReg(NegatedValLo) + .addImm(31) + .setOperandDead(3); // Dead scc + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), Op1L_Op0H_Reg) + .add(Op1L) + .addReg(NegatedHi->getOperand(0).getReg()); + } + Register LowOpcode = Opc == AMDGPU::S_SUB_U64_PSEUDO + ? NegatedValLo + : NewAccumulator->getOperand(0).getReg(); + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), DestSub0) + .add(Op1L) + .addReg(LowOpcode); + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_HI_U32), CarryReg) + .add(Op1L) + .addReg(LowOpcode); + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_MUL_I32), Op1H_Op0L_Reg) + .add(Op1H) + .addReg(LowOpcode); + + Register HiVal = Opc == AMDGPU::S_SUB_U64_PSEUDO ? AddReg : DestSub1; + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_ADD_U32), HiVal) + .addReg(CarryReg) + .addReg(Op1H_Op0L_Reg) + .setOperandDead(3); // Dead scc + + if (Opc == AMDGPU::S_SUB_U64_PSEUDO) { + BuildMI(BB, MI, DL, TII->get(AMDGPU::S_ADD_U32), DestSub1) + .addReg(HiVal) + .addReg(Op1L_Op0H_Reg) + .setOperandDead(3); // Dead scc + } + BuildMI(BB, MI, DL, TII->get(TargetOpcode::REG_SEQUENCE), DstReg) + .addReg(DestSub0) + .addImm(AMDGPU::sub0) + .addReg(DestSub1) + .addImm(AMDGPU::sub1); + break; + } } RetBB = &BB; } @@ -5196,55 +5334,193 @@ static MachineBasicBlock *lowerWaveReduce(MachineInstr &MI, const TargetRegisterClass *WaveMaskRegClass = TRI->getWaveMaskRegClass(); const TargetRegisterClass *DstRegClass = MRI.getRegClass(DstReg); Register LoopIterator = MRI.createVirtualRegister(WaveMaskRegClass); - Register InitalValReg = MRI.createVirtualRegister(DstRegClass); - + Register IdentityValReg = MRI.createVirtualRegister(DstRegClass); Register AccumulatorReg = MRI.createVirtualRegister(DstRegClass); Register ActiveBitsReg = MRI.createVirtualRegister(WaveMaskRegClass); Register NewActiveBitsReg = MRI.createVirtualRegister(WaveMaskRegClass); - - Register FF1Reg = MRI.createVirtualRegister(DstRegClass); - Register LaneValueReg = - MRI.createVirtualRegister(&AMDGPU::SReg_32_XM0RegClass); + Register FF1Reg = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register LaneValueReg = MRI.createVirtualRegister(DstRegClass); bool IsWave32 = ST.isWave32(); - unsigned MovOpc = IsWave32 ? AMDGPU::S_MOV_B32 : AMDGPU::S_MOV_B64; + unsigned MovOpcForExec = IsWave32 ? AMDGPU::S_MOV_B32 : AMDGPU::S_MOV_B64; unsigned ExecReg = IsWave32 ? AMDGPU::EXEC_LO : AMDGPU::EXEC; // Create initial values of induction variable from Exec, Accumulator and // insert branch instr to newly created ComputeBlock - uint32_t InitalValue = getIdentityValueForWaveReduction(Opc); - auto TmpSReg = - BuildMI(BB, I, DL, TII->get(MovOpc), LoopIterator).addReg(ExecReg); - BuildMI(BB, I, DL, TII->get(AMDGPU::S_MOV_B32), InitalValReg) - .addImm(InitalValue); + uint32_t IdentityValue = getIdentityValueForWaveReduction(Opc); + auto TmpSReg = BuildMI(BB, I, DL, TII->get(MovOpcForExec), LoopIterator) + .addReg(ExecReg); + if (is32BitOpc) { + BuildMI(BB, I, DL, TII->get(AMDGPU::S_MOV_B32), IdentityValReg) + .addImm(IdentityValue); + } else { + Register Identitylo = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register Identityhi = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + MachineInstr *IdenHi = + BuildMI(BB, I, DL, TII->get(AMDGPU::S_MOV_B32), Identityhi) + .addImm(IdentityValue); + switch (Opc) { + case AMDGPU::V_CMP_LT_U64_e64: + case AMDGPU::V_CMP_LT_I64_e64: + IdentityValue = int32_t(-1); // u|min + break; + case AMDGPU::V_CMP_GT_U64_e64: + case AMDGPU::V_CMP_GT_I64_e64: + IdentityValue = int32_t(0); // u|max + break; + } + MachineInstr *IdenLo = + BuildMI(BB, I, DL, TII->get(AMDGPU::S_MOV_B32), Identitylo) + .addImm(IdentityValue); + BuildMI(BB, I, DL, TII->get(TargetOpcode::REG_SEQUENCE), IdentityValReg) + .addReg(IdenLo->getOperand(0).getReg()) + .addImm(AMDGPU::sub0) + .addReg(IdenHi->getOperand(0).getReg()) + .addImm(AMDGPU::sub1); + } // clang-format off BuildMI(BB, I, DL, TII->get(AMDGPU::S_BRANCH)) .addMBB(ComputeLoop); // clang-format on // Start constructing ComputeLoop - I = ComputeLoop->end(); + I = ComputeLoop->begin(); auto Accumulator = BuildMI(*ComputeLoop, I, DL, TII->get(AMDGPU::PHI), AccumulatorReg) - .addReg(InitalValReg) + .addReg(IdentityValReg) .addMBB(&BB); auto ActiveBits = BuildMI(*ComputeLoop, I, DL, TII->get(AMDGPU::PHI), ActiveBitsReg) .addReg(TmpSReg->getOperand(0).getReg()) .addMBB(&BB); + I = ComputeLoop->end(); + MachineInstr *NewAccumulator; // Perform the computations unsigned SFFOpc = IsWave32 ? AMDGPU::S_FF1_I32_B32 : AMDGPU::S_FF1_I32_B64; auto FF1 = BuildMI(*ComputeLoop, I, DL, TII->get(SFFOpc), FF1Reg) .addReg(ActiveBits->getOperand(0).getReg()); - auto LaneValue = BuildMI(*ComputeLoop, I, DL, - TII->get(AMDGPU::V_READLANE_B32), LaneValueReg) - .addReg(SrcReg) - .addReg(FF1->getOperand(0).getReg()); - auto NewAccumulator = BuildMI(*ComputeLoop, I, DL, TII->get(Opc), DstReg) - .addReg(Accumulator->getOperand(0).getReg()) - .addReg(LaneValue->getOperand(0).getReg()); + if (is32BitOpc) { + MachineInstr *LaneValue = + BuildMI(*ComputeLoop, I, DL, TII->get(AMDGPU::V_READLANE_B32), + LaneValueReg) + .addReg(SrcReg) + .addReg(FF1->getOperand(0).getReg()); + NewAccumulator = BuildMI(*ComputeLoop, I, DL, TII->get(Opc), DstReg) + .addReg(Accumulator->getOperand(0).getReg()) + .addReg(LaneValue->getOperand(0).getReg()); + } else { + Register LaneValueLoReg = + MRI.createVirtualRegister(&AMDGPU::SReg_32_XM0RegClass); + Register LaneValueHiReg = + MRI.createVirtualRegister(&AMDGPU::SReg_32_XM0RegClass); + Register LaneValReg = MRI.createVirtualRegister(&AMDGPU::SReg_64RegClass); + const TargetRegisterClass *SrcRC = MRI.getRegClass(SrcReg); + const TargetRegisterClass *SrcSubRC = + TRI->getSubRegisterClass(SrcRC, AMDGPU::sub0); + MachineOperand Op1L = TII->buildExtractSubRegOrImm( + MI, MRI, MI.getOperand(1), SrcRC, AMDGPU::sub0, SrcSubRC); + MachineOperand Op1H = TII->buildExtractSubRegOrImm( + MI, MRI, MI.getOperand(1), SrcRC, AMDGPU::sub1, SrcSubRC); + // lane value input should be in an sgpr + MachineInstr *LaneValueLo = + BuildMI(*ComputeLoop, I, DL, TII->get(AMDGPU::V_READLANE_B32), + LaneValueLoReg) + .add(Op1L) + .addReg(FF1->getOperand(0).getReg()); + MachineInstr *LaneValueHi = + BuildMI(*ComputeLoop, I, DL, TII->get(AMDGPU::V_READLANE_B32), + LaneValueHiReg) + .add(Op1H) + .addReg(FF1->getOperand(0).getReg()); + auto LaneValue = BuildMI(*ComputeLoop, I, DL, + TII->get(TargetOpcode::REG_SEQUENCE), LaneValReg) + .addReg(LaneValueLo->getOperand(0).getReg()) + .addImm(AMDGPU::sub0) + .addReg(LaneValueHi->getOperand(0).getReg()) + .addImm(AMDGPU::sub1); + switch (Opc) { + case ::AMDGPU::S_OR_B64: + case ::AMDGPU::S_AND_B64: + case ::AMDGPU::S_XOR_B64: { + NewAccumulator = BuildMI(*ComputeLoop, I, DL, TII->get(Opc), DstReg) + .addReg(Accumulator->getOperand(0).getReg()) + .addReg(LaneValue->getOperand(0).getReg()) + .setOperandDead(3); // Dead scc + break; + } + case AMDGPU::V_CMP_GT_I64_e64: + case AMDGPU::V_CMP_GT_U64_e64: + case AMDGPU::V_CMP_LT_I64_e64: + case AMDGPU::V_CMP_LT_U64_e64: { + Register LaneMaskReg = MRI.createVirtualRegister(WaveMaskRegClass); + Register ComparisonResultReg = + MRI.createVirtualRegister(WaveMaskRegClass); + const TargetRegisterClass *VregClass = + ST.needsAlignedVGPRs() ? &AMDGPU::VReg_64_Align2RegClass + : &AMDGPU::VReg_64RegClass; + const TargetRegisterClass *VSubRegClass = + TRI->getSubRegisterClass(VregClass, AMDGPU::sub0); + Register AccumulatorVReg = MRI.createVirtualRegister(VregClass); + MachineOperand SrcReg0Sub0 = + TII->buildExtractSubRegOrImm(MI, MRI, Accumulator->getOperand(0), + VregClass, AMDGPU::sub0, VSubRegClass); + MachineOperand SrcReg0Sub1 = + TII->buildExtractSubRegOrImm(MI, MRI, Accumulator->getOperand(0), + VregClass, AMDGPU::sub1, VSubRegClass); + BuildMI(*ComputeLoop, I, DL, TII->get(TargetOpcode::REG_SEQUENCE), + AccumulatorVReg) + .add(SrcReg0Sub0) + .addImm(AMDGPU::sub0) + .add(SrcReg0Sub1) + .addImm(AMDGPU::sub1); + auto LaneMask = BuildMI(*ComputeLoop, I, DL, TII->get(Opc), LaneMaskReg) + .addReg(LaneValue->getOperand(0).getReg()) + .addReg(AccumulatorVReg); + + unsigned AndOpc = IsWave32 ? AMDGPU::S_AND_B32 : AMDGPU::S_AND_B64; + BuildMI(*ComputeLoop, I, DL, TII->get(AndOpc), ComparisonResultReg) + .addReg(LaneMask->getOperand(0).getReg()) + .addReg(ActiveBits->getOperand(0).getReg()); + NewAccumulator = BuildMI(*ComputeLoop, I, DL, + TII->get(AMDGPU::S_CSELECT_B64), DstReg) + .addReg(LaneValue->getOperand(0).getReg()) + .addReg(Accumulator->getOperand(0).getReg()); + break; + } + case ::AMDGPU::S_ADD_U64_PSEUDO: + case ::AMDGPU::S_SUB_U64_PSEUDO: { + unsigned newOpc1 = Opc == AMDGPU::S_ADD_U64_PSEUDO ? AMDGPU::S_ADD_U32 + : AMDGPU::S_SUB_U32; + unsigned newOpc2 = Opc == AMDGPU::S_ADD_U64_PSEUDO ? AMDGPU::S_ADDC_U32 + : AMDGPU::S_SUBB_U32; + Register DestLo = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + Register DestHi = MRI.createVirtualRegister(&AMDGPU::SReg_32RegClass); + MachineOperand Accumlo = TII->buildExtractSubRegOrImm( + MI, MRI, Accumulator->getOperand(0), DstRegClass, AMDGPU::sub0, + &AMDGPU::SReg_32RegClass); + MachineOperand Accumhi = TII->buildExtractSubRegOrImm( + MI, MRI, Accumulator->getOperand(0), DstRegClass, AMDGPU::sub1, + &AMDGPU::SReg_32RegClass); + MachineInstr *DestLoComputation = + BuildMI(*ComputeLoop, I, DL, TII->get(newOpc1), DestLo) + .add(Accumlo) + .addReg(LaneValueLo->getOperand(0).getReg()); + MachineInstr *DestHiComputation = + BuildMI(*ComputeLoop, I, DL, TII->get(newOpc2), DestHi) + .add(Accumhi) + .addReg(LaneValueHi->getOperand(0).getReg()); + NewAccumulator = BuildMI(*ComputeLoop, I, DL, + TII->get(TargetOpcode::REG_SEQUENCE), DstReg) + .addReg(DestLoComputation->getOperand(0).getReg()) + .addImm(AMDGPU::sub0) + .addReg(DestHiComputation->getOperand(0).getReg()) + .addImm(AMDGPU::sub1); + break; + } + } + } // Manipulate the iterator to get the next active lane unsigned BITSETOpc = IsWave32 ? AMDGPU::S_BITSET0_B32 : AMDGPU::S_BITSET0_B64; @@ -5284,22 +5560,40 @@ SITargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI, switch (MI.getOpcode()) { case AMDGPU::WAVE_REDUCE_UMIN_PSEUDO_U32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_MIN_U32); + case AMDGPU::WAVE_REDUCE_UMIN_PSEUDO_U64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::V_CMP_LT_U64_e64); case AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_MIN_I32); + case AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::V_CMP_LT_I64_e64); case AMDGPU::WAVE_REDUCE_UMAX_PSEUDO_U32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_MAX_U32); + case AMDGPU::WAVE_REDUCE_UMAX_PSEUDO_U64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::V_CMP_GT_U64_e64); case AMDGPU::WAVE_REDUCE_MAX_PSEUDO_I32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_MAX_I32); + case AMDGPU::WAVE_REDUCE_MAX_PSEUDO_I64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::V_CMP_GT_I64_e64); case AMDGPU::WAVE_REDUCE_ADD_PSEUDO_I32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_ADD_I32); + case AMDGPU::WAVE_REDUCE_ADD_PSEUDO_I64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_ADD_U64_PSEUDO); case AMDGPU::WAVE_REDUCE_SUB_PSEUDO_I32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_SUB_I32); + case AMDGPU::WAVE_REDUCE_SUB_PSEUDO_I64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_SUB_U64_PSEUDO); case AMDGPU::WAVE_REDUCE_AND_PSEUDO_B32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_AND_B32); + case AMDGPU::WAVE_REDUCE_AND_PSEUDO_B64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_AND_B64); case AMDGPU::WAVE_REDUCE_OR_PSEUDO_B32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_OR_B32); + case AMDGPU::WAVE_REDUCE_OR_PSEUDO_B64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_OR_B64); case AMDGPU::WAVE_REDUCE_XOR_PSEUDO_B32: return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_XOR_B32); + case AMDGPU::WAVE_REDUCE_XOR_PSEUDO_B64: + return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_XOR_B64); case AMDGPU::S_UADDO_PSEUDO: case AMDGPU::S_USUBO_PSEUDO: { const DebugLoc &DL = MI.getDebugLoc(); diff --git a/llvm/lib/Target/AMDGPU/SIInstructions.td b/llvm/lib/Target/AMDGPU/SIInstructions.td index 4419ce00b473c..142608fa536cc 100644 --- a/llvm/lib/Target/AMDGPU/SIInstructions.td +++ b/llvm/lib/Target/AMDGPU/SIInstructions.td @@ -304,28 +304,57 @@ def : GCNPat<(i32 (int_amdgcn_set_inactive_chain_arg i32:$src, i32:$inactive)), (V_SET_INACTIVE_B32 0, VGPR_32:$src, 0, VGPR_32:$inactive, (IMPLICIT_DEF))>; // clang-format off -defvar int_amdgcn_wave_reduce_ = "int_amdgcn_wave_reduce_"; + multiclass - AMDGPUWaveReducePseudoGenerator { + AMDGPUWaveReducePseudoGenerator { let usesCustomInserter = 1, hasSideEffects = 0, mayLoad = 0, mayStore = 0, Uses = [EXEC] in { def !toupper(Op) #"_PSEUDO_" #DataType - : VPseudoInstSI<(outs SGPR_32 : $sdst), - (ins VSrc_b32 : $src, VSrc_b32 : $strategy), - [(set i32 : $sdst, (!cast(int_amdgcn_wave_reduce_ #Op) i32 : $src, i32 : $strategy))]> {} + : VPseudoInstSI<(outs RetReg : $sdst), + (ins Reg : $src, VSrc_b32 : $strategy), + [(set ty : $sdst, (!cast("int_amdgcn_wave_reduce_" #Op) ty : $src, i32 : $strategy))]> {} } } // clang-format on +class WaveReduceOp { + string Name = OpName; + string TypeString = TypeStr; + ValueType VT = Ty; + RegisterClass RetReg = ReturnRegisterClass; + SrcRegOrImm9 Reg = RC; +} + // Input list : [Operation_name, -// type - Signed(I)/Unsigned(U)/Float(F)/Bitwise(B)] +// type - Signed(I)/Unsigned(U)/Float(F)/Bitwise(B), +// bit-width +// output register class, +// input register class] defvar Operations = [ - ["umin", "U32"], ["min", "I32"], ["umax", "U32"], ["max", "I32"], - ["add", "I32"], ["sub", "I32"], ["and", "B32"], ["or", "B32"], - ["xor", "B32"] + WaveReduceOp<"umin", "U32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"min", "I32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"umax", "U32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"max", "I32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"add", "I32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"sub", "I32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"and", "B32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"or", "B32", i32, SGPR_32, VSrc_b32>, + WaveReduceOp<"xor", "B32", i32, SGPR_32, VSrc_b32>, + + WaveReduceOp<"umin", "U64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"min", "I64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"umax", "U64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"max", "I64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"add", "I64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"sub", "I64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"and", "B64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"or", "B64", i64, SGPR_64, VSrc_b64>, + WaveReduceOp<"xor", "B64", i64, SGPR_64, VSrc_b64>, ]; foreach Op = Operations in { - defm WAVE_REDUCE_ : AMDGPUWaveReducePseudoGenerator; + defm WAVE_REDUCE_ : AMDGPUWaveReducePseudoGenerator; } let usesCustomInserter = 1, Defs = [VCC] in { diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll index d2ca1d8136043..b6af8b4bb798d 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll @@ -1226,6 +1226,1362 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: s_mul_i32 s0, s2, s4 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s1, s2, s4 +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s3, s4 +; GFX8DAGISEL-NEXT: s_add_u32 s1, s1, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s5, s[4:5] +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX8GISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s5, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s5, s[4:5] +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9GISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: uniform_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: uniform_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1064GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: uniform_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: uniform_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1032GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1164GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1164GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1132GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.add.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s3 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX8GISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX8GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s3 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX8GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX9DAGISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s3 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX9GISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX9GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s3 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX9GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: const_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1064DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: const_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064GISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1064GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1064GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: const_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1032DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: const_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1032GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1032GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1164DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1164GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1164GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1132DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_3) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s4, s2, 0 +; GFX1132GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1132GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.add.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s3 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX8GISEL-NEXT: s_mul_hi_u32 s4, s0, s3 +; GFX8GISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX8GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s3 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s3, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX9GISEL-NEXT: s_mul_hi_u32 s4, s0, s3 +; GFX9GISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX9GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: poison_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: poison_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1064GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: poison_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: poison_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1032GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: poison_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: poison_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1164GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: poison_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: poison_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s3, s0, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s4, s1, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1132GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.add.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX8DAGISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX8DAGISEL-NEXT: s_add_u32 s4, s4, s9 +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX8DAGISEL-NEXT: s_addc_u32 s5, s5, s10 +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b32 s4, 0 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_mov_b32 s5, s4 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX8GISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX8GISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX8GISEL-NEXT: s_add_u32 s4, s4, s9 +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX8GISEL-NEXT: s_addc_u32 s5, s5, s10 +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX9DAGISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX9DAGISEL-NEXT: s_add_u32 s4, s4, s9 +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX9DAGISEL-NEXT: s_addc_u32 s5, s5, s10 +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b32 s4, 0 +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_mov_b32 s5, s4 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX9GISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX9GISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX9GISEL-NEXT: s_add_u32 s4, s4, s9 +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX9GISEL-NEXT: s_addc_u32 s5, s5, s10 +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX1064DAGISEL-NEXT: s_add_u32 s4, s4, s9 +; GFX1064DAGISEL-NEXT: s_addc_u32 s5, s5, s10 +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX1064GISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX1064GISEL-NEXT: s_add_u32 s4, s4, s9 +; GFX1064GISEL-NEXT: s_addc_u32 s5, s5, s10 +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_add_u32 s4, s4, s8 +; GFX1032DAGISEL-NEXT: s_addc_u32 s5, s5, s9 +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_add_u32 s4, s4, s8 +; GFX1032GISEL-NEXT: s_addc_u32 s5, s5, s9 +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s4, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v2, s4 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s6, v3, s4 +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s4 +; GFX1164DAGISEL-NEXT: s_add_u32 s0, s0, s5 +; GFX1164DAGISEL-NEXT: s_addc_u32 s1, s1, s6 +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s4, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v2, s4 +; GFX1164GISEL-NEXT: v_readlane_b32 s6, v3, s4 +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s4 +; GFX1164GISEL-NEXT: s_add_u32 s0, s0, s5 +; GFX1164GISEL-NEXT: s_addc_u32 s1, s1, s6 +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_add_u32 s0, s0, s4 +; GFX1132DAGISEL-NEXT: s_addc_u32 s1, s1, s5 +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_add_u32 s0, s0, s4 +; GFX1132GISEL-NEXT: s_addc_u32 s1, s1, s5 +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.add.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8DAGISEL-NEXT: ; %bb.1: ; %else +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX8DAGISEL-NEXT: s_add_u32 s7, s2, s3 +; GFX8DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX8DAGISEL-NEXT: ; %bb.3: ; %if +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7] +; GFX8DAGISEL-NEXT: s_mul_i32 s6, s4, s7 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s4, s4, s7 +; GFX8DAGISEL-NEXT: s_mul_i32 s5, s5, s7 +; GFX8DAGISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7] +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX8GISEL-NEXT: s_add_u32 s7, s2, s3 +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7] +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s6, s4, s7 +; GFX8GISEL-NEXT: s_mul_hi_u32 s4, s4, s7 +; GFX8GISEL-NEXT: s_mul_i32 s5, s5, s7 +; GFX8GISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9DAGISEL-NEXT: ; %bb.1: ; %else +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s5, s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX9DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX9DAGISEL-NEXT: ; %bb.3: ; %if +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s5, s[4:5] +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s6, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s6, s6, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s5, s7, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s6, s5 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7] +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX9GISEL-NEXT: s_add_u32 s7, s2, s3 +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[8:9], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s6, s8, s4 +; GFX9GISEL-NEXT: s_mul_hi_u32 s5, s8, s4 +; GFX9GISEL-NEXT: s_mul_i32 s4, s9, s4 +; GFX9GISEL-NEXT: s_add_u32 s7, s5, s4 +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1064DAGISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s9, s2, s8 +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s8 +; GFX1064DAGISEL-NEXT: s_mul_i32 s8, s2, s8 +; GFX1064DAGISEL-NEXT: s_add_u32 s9, s9, s3 +; GFX1064DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s8 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s9 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX1064DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s5, s6, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s7, s7, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s6, s4 +; GFX1064DAGISEL-NEXT: s_add_u32 s5, s5, s7 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1064DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1064GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1064GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s5, s6, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s7, s7, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s6, s6, s4 +; GFX1064GISEL-NEXT: s_add_u32 s7, s5, s7 +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s2, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s5, s3 +; GFX1032DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s2, s8 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2 +; GFX1032DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s5, s6, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s7, s7, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s6, s3 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s5, s7 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1032DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1032GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s4, s6, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s5, s7, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s6, s3 +; GFX1032GISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1164DAGISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s9, s2, s8 +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s8 +; GFX1164DAGISEL-NEXT: s_mul_i32 s8, s2, s8 +; GFX1164DAGISEL-NEXT: s_add_u32 s9, s9, s3 +; GFX1164DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s8 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s9 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX1164DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s7, s4, s6 +; GFX1164DAGISEL-NEXT: s_mul_i32 s5, s5, s6 +; GFX1164DAGISEL-NEXT: s_mul_i32 s4, s4, s6 +; GFX1164DAGISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1164GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s7, s4, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s5, s5, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s6, s4, s6 +; GFX1164GISEL-NEXT: s_add_u32 s7, s7, s5 +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1132DAGISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1132DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s2, s8 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2 +; GFX1132DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_3) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s6, s4, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s5, s5, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s4, s4, s3 +; GFX1132DAGISEL-NEXT: s_add_u32 s5, s6, s5 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1132GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s7, s4, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s5, s5, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s4, s3 +; GFX1132GISEL-NEXT: s_add_u32 s7, s7, s5 +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.add.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.add.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: ; GFX10DAGISEL: {{.*}} ; GFX10GISEL: {{.*}} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.and.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.and.ll index 356b0e73b39e7..55e6189f65675 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.and.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.and.ll @@ -980,3 +980,857 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: uniform_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: uniform_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.and.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: const_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: const_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132GISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.and.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: poison_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: poison_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX11DAGISEL-LABEL: poison_value_i64: +; GFX11DAGISEL: ; %bb.0: ; %entry +; GFX11DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11DAGISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11DAGISEL-NEXT: s_endpgm +; +; GFX11GISEL-LABEL: poison_value_i64: +; GFX11GISEL: ; %bb.0: ; %entry +; GFX11GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11GISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.and.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX8DAGISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b32 s4, -1 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_mov_b32 s5, s4 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX8GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX8GISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX9DAGISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b32 s4, -1 +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_mov_b32 s5, s4 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX9GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX9GISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX1064DAGISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b32 s4, -1 +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX1064GISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s4, -1 +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_and_b64 s[4:5], s[4:5], s[8:9] +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, -1 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s6, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s6 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v3, s6 +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s6 +; GFX1164DAGISEL-NEXT: s_and_b64 s[0:1], s[0:1], s[4:5] +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b32 s0, -1 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s6, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s6 +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v3, s6 +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s6 +; GFX1164GISEL-NEXT: s_and_b64 s[0:1], s[0:1], s[4:5] +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, -1 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_and_b64 s[0:1], s[0:1], s[4:5] +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s0, -1 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_and_b64 s[0:1], s[0:1], s[4:5] +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.and.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: ; %bb.1: ; %if +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: ; %bb.1: ; %if +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s4, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s4, exec_lo, s4 +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s6, exec_lo, s6 +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.and.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.and.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.max.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.max.ll index 7dc0cb05b0abe..96b67e71fcd28 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.max.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.max.ll @@ -980,3 +980,903 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: uniform_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: uniform_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.max.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: const_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: const_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132GISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.max.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: poison_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: poison_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX11DAGISEL-LABEL: poison_value_i64: +; GFX11DAGISEL: ; %bb.0: ; %entry +; GFX11DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11DAGISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11DAGISEL-NEXT: s_endpgm +; +; GFX11GISEL-LABEL: poison_value_i64: +; GFX11GISEL: ; %bb.0: ; %entry +; GFX11GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11GISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.max.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_brev_b32 s5, 1 +; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX8DAGISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[8:9], v[4:5] +; GFX8DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX8DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_brev_b32 s5, 1 +; GFX8GISEL-NEXT: s_mov_b32 s4, 0 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX8GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX8GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX8GISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[8:9], v[4:5] +; GFX8GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX8GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_brev_b32 s5, 1 +; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX9DAGISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[8:9], v[4:5] +; GFX9DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX9DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_brev_b32 s5, 1 +; GFX9GISEL-NEXT: s_mov_b32 s4, 0 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX9GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX9GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX9GISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[8:9], v[4:5] +; GFX9GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX9GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_brev_b32 s5, 1 +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX1064DAGISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[8:9], v[4:5] +; GFX1064DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX1064DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_brev_b32 s5, 1 +; GFX1064GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX1064GISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[8:9], v[4:5] +; GFX1064GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX1064GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_brev_b32 s5, 1 +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: v_cmp_gt_i64_e32 vcc_lo, s[8:9], v[4:5] +; GFX1032DAGISEL-NEXT: s_and_b32 s10, vcc_lo, s6 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_brev_b32 s5, 1 +; GFX1032GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: v_cmp_gt_i64_e32 vcc_lo, s[8:9], v[4:5] +; GFX1032GISEL-NEXT: s_and_b32 s10, vcc_lo, s6 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_brev_b32 s1, 1 +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s8, s[2:3] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v4, s0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v5, s1 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s8 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v3, s8 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164DAGISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[4:5], v[4:5] +; GFX1164DAGISEL-NEXT: s_and_b64 s[6:7], vcc, s[2:3] +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s8 +; GFX1164DAGISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_brev_b32 s1, 1 +; GFX1164GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s8, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v4, s0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v5, s1 +; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s8 +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v3, s8 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmp_gt_i64_e32 vcc, s[4:5], v[4:5] +; GFX1164GISEL-NEXT: s_and_b64 s[6:7], vcc, s[2:3] +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s8 +; GFX1164GISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_brev_b32 s1, 1 +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132DAGISEL-NEXT: v_cmp_gt_i64_e32 vcc_lo, s[4:5], v[4:5] +; GFX1132DAGISEL-NEXT: s_and_b32 s6, vcc_lo, s2 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_brev_b32 s1, 1 +; GFX1132GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmp_gt_i64_e32 vcc_lo, s[4:5], v[4:5] +; GFX1132GISEL-NEXT: s_and_b32 s6, vcc_lo, s2 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.max.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: ; %bb.1: ; %if +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: ; %bb.1: ; %if +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s4, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s4, exec_lo, s4 +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s6, exec_lo, s6 +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.max.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.max.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.min.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.min.ll index 7cb0e6533c722..4e0c9ec111cbe 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.min.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.min.ll @@ -980,3 +980,903 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: uniform_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: uniform_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.min.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: const_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: const_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132GISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.min.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: poison_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: poison_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX11DAGISEL-LABEL: poison_value_i64: +; GFX11DAGISEL: ; %bb.0: ; %entry +; GFX11DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11DAGISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11DAGISEL-NEXT: s_endpgm +; +; GFX11GISEL-LABEL: poison_value_i64: +; GFX11GISEL: ; %bb.0: ; %entry +; GFX11GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11GISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.min.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_brev_b32 s5, -2 +; GFX8DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX8DAGISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[8:9], v[4:5] +; GFX8DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX8DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_brev_b32 s5, -2 +; GFX8GISEL-NEXT: s_mov_b32 s4, -1 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX8GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX8GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX8GISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[8:9], v[4:5] +; GFX8GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX8GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_brev_b32 s5, -2 +; GFX9DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX9DAGISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[8:9], v[4:5] +; GFX9DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX9DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_brev_b32 s5, -2 +; GFX9GISEL-NEXT: s_mov_b32 s4, -1 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX9GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX9GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX9GISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[8:9], v[4:5] +; GFX9GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX9GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_brev_b32 s5, -2 +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX1064DAGISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[8:9], v[4:5] +; GFX1064DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX1064DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_brev_b32 s5, -2 +; GFX1064GISEL-NEXT: s_mov_b32 s4, -1 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX1064GISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[8:9], v[4:5] +; GFX1064GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX1064GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_brev_b32 s5, -2 +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: v_cmp_lt_i64_e32 vcc_lo, s[8:9], v[4:5] +; GFX1032DAGISEL-NEXT: s_and_b32 s10, vcc_lo, s6 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_brev_b32 s5, -2 +; GFX1032GISEL-NEXT: s_mov_b32 s4, -1 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: v_cmp_lt_i64_e32 vcc_lo, s[8:9], v[4:5] +; GFX1032GISEL-NEXT: s_and_b32 s10, vcc_lo, s6 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_brev_b32 s1, -2 +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, -1 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s8, s[2:3] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v4, s0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v5, s1 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s8 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v3, s8 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164DAGISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[4:5], v[4:5] +; GFX1164DAGISEL-NEXT: s_and_b64 s[6:7], vcc, s[2:3] +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s8 +; GFX1164DAGISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_brev_b32 s1, -2 +; GFX1164GISEL-NEXT: s_mov_b32 s0, -1 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s8, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v4, s0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v5, s1 +; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s8 +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v3, s8 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmp_lt_i64_e32 vcc, s[4:5], v[4:5] +; GFX1164GISEL-NEXT: s_and_b64 s[6:7], vcc, s[2:3] +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s8 +; GFX1164GISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_brev_b32 s1, -2 +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, -1 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132DAGISEL-NEXT: v_cmp_lt_i64_e32 vcc_lo, s[4:5], v[4:5] +; GFX1132DAGISEL-NEXT: s_and_b32 s6, vcc_lo, s2 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_brev_b32 s1, -2 +; GFX1132GISEL-NEXT: s_mov_b32 s0, -1 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmp_lt_i64_e32 vcc_lo, s[4:5], v[4:5] +; GFX1132GISEL-NEXT: s_and_b32 s6, vcc_lo, s2 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.min.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: ; %bb.1: ; %if +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: ; %bb.1: ; %if +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s4, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s4, exec_lo, s4 +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s6, exec_lo, s6 +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.min.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.min.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.or.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.or.ll index e08787e6ba70a..5dcd1de6d05d7 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.or.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.or.ll @@ -12,6 +12,8 @@ ; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 < %s | FileCheck -check-prefixes=GFX11DAGISEL,GFX1132DAGISEL %s ; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 < %s | FileCheck -check-prefixes=GFX11GISEL,GFX1132GISEL %s +declare i32 @llvm.amdgcn.workitem.id.x() #0 + define amdgpu_kernel void @uniform_value(ptr addrspace(1) %out, i32 %in) { ; GFX8DAGISEL-LABEL: uniform_value: ; GFX8DAGISEL: ; %bb.0: ; %entry @@ -980,3 +982,858 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: uniform_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: uniform_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.or.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: const_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: const_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132GISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.or.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: poison_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: poison_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v0, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX11DAGISEL-LABEL: poison_value_i64: +; GFX11DAGISEL: ; %bb.0: ; %entry +; GFX11DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11DAGISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11DAGISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11DAGISEL-NEXT: s_endpgm +; +; GFX11GISEL-LABEL: poison_value_i64: +; GFX11GISEL: ; %bb.0: ; %entry +; GFX11GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX11GISEL-NEXT: v_mov_b32_e32 v0, 0 +; GFX11GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX11GISEL-NEXT: global_store_b64 v0, v[0:1], s[0:1] +; GFX11GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.or.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX8DAGISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b32 s4, 0 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_mov_b32 s5, s4 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX8GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX8GISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX9DAGISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b32 s4, 0 +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_mov_b32 s5, s4 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX9GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX9GISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX1064DAGISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX1064GISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_or_b64 s[4:5], s[4:5], s[8:9] +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s6, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s6 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v3, s6 +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s6 +; GFX1164DAGISEL-NEXT: s_or_b64 s[0:1], s[0:1], s[4:5] +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s6, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s6 +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v3, s6 +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s6 +; GFX1164GISEL-NEXT: s_or_b64 s[0:1], s[0:1], s[4:5] +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_or_b64 s[0:1], s[0:1], s[4:5] +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_or_b64 s[0:1], s[0:1], s[4:5] +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + ; %id.x = call i32 @llvm.amdgcn.workitem.id.x() + %result = call i64 @llvm.amdgcn.wave.reduce.or.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: ; %bb.1: ; %if +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: ; %bb.1: ; %if +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s4, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s4, exec_lo, s4 +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s6, exec_lo, s6 +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.or.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.or.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll index edb888a21f735..aff5c01ee174b 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll @@ -1275,6 +1275,1621 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX8DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: s_ashr_i32 s0, s4, 31 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: s_mul_i32 s1, s2, s0 +; GFX8DAGISEL-NEXT: s_mul_i32 s0, s2, s4 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s4 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX8DAGISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s1, s2, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX8GISEL-NEXT: s_mul_i32 s5, s4, -1 +; GFX8GISEL-NEXT: s_ashr_i32 s4, s5, 31 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s6, s2, s4 +; GFX8GISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX8GISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX8GISEL-NEXT: s_add_u32 s5, s2, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9DAGISEL-NEXT: s_mul_i32 s5, s4, -1 +; GFX9DAGISEL-NEXT: s_ashr_i32 s4, s5, 31 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s6, s2, s4 +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9GISEL-NEXT: s_mul_i32 s5, s4, -1 +; GFX9GISEL-NEXT: s_ashr_i32 s4, s5, 31 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s6, s2, s4 +; GFX9GISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9GISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX9GISEL-NEXT: s_add_u32 s5, s2, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: uniform_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1064DAGISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: uniform_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064GISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1064GISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1064GISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1064GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1064GISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: uniform_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1032DAGISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: uniform_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032GISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1032GISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032GISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1032GISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1032GISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1164DAGISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1164DAGISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1164GISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1164GISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1164GISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1164GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1164GISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1132DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1132DAGISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1132GISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s6, s2, s4 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1132GISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1132GISEL-NEXT: s_add_u32 s3, s6, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1132GISEL-NEXT: s_add_u32 s3, s3, s5 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.sub.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX8DAGISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX8DAGISEL-NEXT: s_mul_i32 s4, s2, 0x7b +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s5, 0x7b, s3 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8GISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX8GISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX8GISEL-NEXT: s_mul_i32 s4, s2, 0x7b +; GFX8GISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX8GISEL-NEXT: s_mul_hi_u32 s5, 0x7b, s3 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX8GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX8GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, 0x7b +; GFX9DAGISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s5, 0x7b, s3 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9GISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX9GISEL-NEXT: s_mul_i32 s4, s2, 0x7b +; GFX9GISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX9GISEL-NEXT: s_mul_hi_u32 s5, 0x7b, s3 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, 0 +; GFX9GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX9GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: const_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1064DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1064DAGISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1064DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1064DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: const_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1064GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1064GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1064GISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1064GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1064GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1064GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: const_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1032DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1032DAGISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1032DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1032DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: const_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1032GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1032GISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1032GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1032GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1032GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1164DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1164DAGISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1164DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1164DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1164GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1164GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1164GISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1164GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1164GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1164GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1132DAGISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1132DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1132DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1132GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1132GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s5, s2, 0 +; GFX1132GISEL-NEXT: s_mulk_i32 s3, 0x7b +; GFX1132GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1132GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1132GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.sub.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX8DAGISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mul_i32 s4, s0, s2 +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s5, s0, s3 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8GISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX8GISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s4, s0, s2 +; GFX8GISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX8GISEL-NEXT: s_mul_hi_u32 s5, s0, s3 +; GFX8GISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX8GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX8GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX9DAGISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s0, s2 +; GFX9DAGISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s5, s0, s3 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9GISEL-NEXT: s_mul_i32 s3, s2, -1 +; GFX9GISEL-NEXT: s_ashr_i32 s2, s3, 31 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s4, s0, s2 +; GFX9GISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX9GISEL-NEXT: s_mul_hi_u32 s5, s0, s3 +; GFX9GISEL-NEXT: s_mul_i32 s3, s1, s3 +; GFX9GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX9GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: poison_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1064DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1064DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: poison_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1064GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1064GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1064GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1064GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: poison_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1032DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1032DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: poison_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1032GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1032GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1032GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: poison_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1164DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1164DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: poison_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1164GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1164GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1164GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1164GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: poison_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1132DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: poison_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, -1 +; GFX1132GISEL-NEXT: s_ashr_i32 s3, s2, 31 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s5, s1, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s0, s3 +; GFX1132GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX1132GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1132GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.sub.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX8DAGISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX8DAGISEL-NEXT: s_sub_u32 s4, s4, s9 +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX8DAGISEL-NEXT: s_subb_u32 s5, s5, s10 +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b32 s4, 0 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_mov_b32 s5, s4 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX8GISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX8GISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX8GISEL-NEXT: s_sub_u32 s4, s4, s9 +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX8GISEL-NEXT: s_subb_u32 s5, s5, s10 +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX9DAGISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX9DAGISEL-NEXT: s_sub_u32 s4, s4, s9 +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX9DAGISEL-NEXT: s_subb_u32 s5, s5, s10 +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b32 s4, 0 +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_mov_b32 s5, s4 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX9GISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX9GISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX9GISEL-NEXT: s_sub_u32 s4, s4, s9 +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX9GISEL-NEXT: s_subb_u32 s5, s5, s10 +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX1064DAGISEL-NEXT: s_sub_u32 s4, s4, s9 +; GFX1064DAGISEL-NEXT: s_subb_u32 s5, s5, s10 +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7] +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v2, s8 +; GFX1064GISEL-NEXT: v_readlane_b32 s10, v3, s8 +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s8 +; GFX1064GISEL-NEXT: s_sub_u32 s4, s4, s9 +; GFX1064GISEL-NEXT: s_subb_u32 s5, s5, s10 +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_sub_u32 s4, s4, s8 +; GFX1032DAGISEL-NEXT: s_subb_u32 s5, s5, s9 +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_sub_u32 s4, s4, s8 +; GFX1032GISEL-NEXT: s_subb_u32 s5, s5, s9 +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s4, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v2, s4 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s6, v3, s4 +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s4 +; GFX1164DAGISEL-NEXT: s_sub_u32 s0, s0, s5 +; GFX1164DAGISEL-NEXT: s_subb_u32 s1, s1, s6 +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s4, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v2, s4 +; GFX1164GISEL-NEXT: v_readlane_b32 s6, v3, s4 +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s4 +; GFX1164GISEL-NEXT: s_sub_u32 s0, s0, s5 +; GFX1164GISEL-NEXT: s_subb_u32 s1, s1, s6 +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_sub_u32 s0, s0, s4 +; GFX1132DAGISEL-NEXT: s_subb_u32 s1, s1, s5 +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_sub_u32 s0, s0, s4 +; GFX1132GISEL-NEXT: s_subb_u32 s1, s1, s5 +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.sub.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8DAGISEL-NEXT: ; %bb.1: ; %else +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8DAGISEL-NEXT: s_mul_i32 s7, s6, -1 +; GFX8DAGISEL-NEXT: s_ashr_i32 s6, s7, 31 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mul_i32 s10, s2, s6 +; GFX8DAGISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX8DAGISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s7, s2, s10 +; GFX8DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8DAGISEL-NEXT: ; %bb.3: ; %if +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8DAGISEL-NEXT: s_mul_i32 s7, s6, -1 +; GFX8DAGISEL-NEXT: s_ashr_i32 s6, s7, 31 +; GFX8DAGISEL-NEXT: s_mul_i32 s8, s4, s6 +; GFX8DAGISEL-NEXT: s_mul_i32 s6, s4, s7 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s4, s4, s7 +; GFX8DAGISEL-NEXT: s_mul_i32 s5, s5, s7 +; GFX8DAGISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX8DAGISEL-NEXT: s_add_u32 s7, s4, s8 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8DAGISEL-NEXT: .LBB9_4: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8GISEL-NEXT: s_mul_i32 s7, s6, -1 +; GFX8GISEL-NEXT: s_ashr_i32 s6, s7, 31 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s10, s2, s6 +; GFX8GISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX8GISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX8GISEL-NEXT: s_add_u32 s7, s2, s10 +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8GISEL-NEXT: s_mul_i32 s7, s6, -1 +; GFX8GISEL-NEXT: s_ashr_i32 s6, s7, 31 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s8, s4, s6 +; GFX8GISEL-NEXT: s_mul_i32 s6, s4, s7 +; GFX8GISEL-NEXT: s_mul_hi_u32 s4, s4, s7 +; GFX8GISEL-NEXT: s_mul_i32 s5, s5, s7 +; GFX8GISEL-NEXT: s_add_u32 s4, s4, s5 +; GFX8GISEL-NEXT: s_add_u32 s7, s4, s8 +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9DAGISEL-NEXT: ; %bb.1: ; %else +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9DAGISEL-NEXT: s_mul_i32 s5, s4, -1 +; GFX9DAGISEL-NEXT: s_ashr_i32 s4, s5, 31 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s10, s2, s4 +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s10 +; GFX9DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9DAGISEL-NEXT: ; %bb.3: ; %if +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9DAGISEL-NEXT: s_mul_i32 s5, s4, -1 +; GFX9DAGISEL-NEXT: s_ashr_i32 s4, s5, 31 +; GFX9DAGISEL-NEXT: s_mul_i32 s8, s6, s4 +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s6, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s6, s6, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s5, s7, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s6, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s5, s8 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: .LBB9_4: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX9GISEL-NEXT: s_mul_i32 s7, s6, -1 +; GFX9GISEL-NEXT: s_ashr_i32 s6, s7, 31 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s10, s2, s6 +; GFX9GISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX9GISEL-NEXT: s_add_u32 s2, s2, s3 +; GFX9GISEL-NEXT: s_add_u32 s7, s2, s10 +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[8:9], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9GISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX9GISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s6, s8, s4 +; GFX9GISEL-NEXT: s_mul_hi_u32 s7, s8, s4 +; GFX9GISEL-NEXT: s_mul_i32 s4, s9, s4 +; GFX9GISEL-NEXT: s_mul_i32 s5, s8, s5 +; GFX9GISEL-NEXT: s_add_u32 s4, s7, s4 +; GFX9GISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1064DAGISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9] +; GFX1064DAGISEL-NEXT: s_mul_i32 s8, s8, -1 +; GFX1064DAGISEL-NEXT: s_ashr_i32 s9, s8, 31 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s10, s2, s8 +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s8 +; GFX1064DAGISEL-NEXT: s_mul_i32 s9, s2, s9 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s10, s3 +; GFX1064DAGISEL-NEXT: s_mul_i32 s8, s2, s8 +; GFX1064DAGISEL-NEXT: s_add_u32 s9, s3, s9 +; GFX1064DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s8 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s9 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX1064DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1064DAGISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s8, s6, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s7, s7, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s5, s6, s5 +; GFX1064DAGISEL-NEXT: s_add_u32 s7, s8, s7 +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s6, s4 +; GFX1064DAGISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1064DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1064GISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1064GISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s10, s2, s6 +; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1064GISEL-NEXT: s_mul_i32 s7, s2, s7 +; GFX1064GISEL-NEXT: s_add_u32 s3, s10, s3 +; GFX1064GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1064GISEL-NEXT: s_add_u32 s7, s3, s7 +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064GISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1064GISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_hi_u32 s8, s6, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s7, s7, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s5, s6, s5 +; GFX1064GISEL-NEXT: s_add_u32 s7, s8, s7 +; GFX1064GISEL-NEXT: s_mul_i32 s6, s6, s4 +; GFX1064GISEL-NEXT: s_add_u32 s7, s7, s5 +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s4, -1 +; GFX1032DAGISEL-NEXT: s_ashr_i32 s5, s4, 31 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s9, s2, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s5, s2, s5 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s9, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s2, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s3, s5 +; GFX1032DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s2, s8 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2 +; GFX1032DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, -1 +; GFX1032DAGISEL-NEXT: s_ashr_i32 s4, s3, 31 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s5, s6, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s7, s7, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s8, s6, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s5, s7 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s6, s3 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s5, s8 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1032DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1032GISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s9, s2, s6 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1032GISEL-NEXT: s_mul_i32 s7, s2, s7 +; GFX1032GISEL-NEXT: s_add_u32 s3, s9, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1032GISEL-NEXT: s_add_u32 s7, s3, s7 +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, -1 +; GFX1032GISEL-NEXT: s_ashr_i32 s4, s3, 31 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_hi_u32 s5, s6, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s7, s7, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s4, s6, s4 +; GFX1032GISEL-NEXT: s_add_u32 s5, s5, s7 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s6, s3 +; GFX1032GISEL-NEXT: s_add_u32 s7, s5, s4 +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1164DAGISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9] +; GFX1164DAGISEL-NEXT: s_mul_i32 s8, s8, -1 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_ashr_i32 s9, s8, 31 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s10, s2, s8 +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s8 +; GFX1164DAGISEL-NEXT: s_mul_i32 s9, s2, s9 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s10, s3 +; GFX1164DAGISEL-NEXT: s_mul_i32 s8, s2, s8 +; GFX1164DAGISEL-NEXT: s_add_u32 s9, s3, s9 +; GFX1164DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s8 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s9 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX1164DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1164DAGISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s8, s4, s6 +; GFX1164DAGISEL-NEXT: s_mul_i32 s5, s5, s6 +; GFX1164DAGISEL-NEXT: s_mul_i32 s7, s4, s7 +; GFX1164DAGISEL-NEXT: s_add_u32 s5, s8, s5 +; GFX1164DAGISEL-NEXT: s_mul_i32 s4, s4, s6 +; GFX1164DAGISEL-NEXT: s_add_u32 s5, s5, s7 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164GISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s10, s2, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s7, s2, s7 +; GFX1164GISEL-NEXT: s_add_u32 s3, s10, s3 +; GFX1164GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1164GISEL-NEXT: s_add_u32 s7, s3, s7 +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164GISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_hi_u32 s8, s4, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s5, s5, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s7, s4, s7 +; GFX1164GISEL-NEXT: s_add_u32 s5, s8, s5 +; GFX1164GISEL-NEXT: s_mul_i32 s6, s4, s6 +; GFX1164GISEL-NEXT: s_add_u32 s7, s5, s7 +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s9, s2, s6 +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1132DAGISEL-NEXT: s_mul_i32 s7, s2, s7 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s9, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1132DAGISEL-NEXT: s_add_u32 s7, s3, s7 +; GFX1132DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s2, s8 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2 +; GFX1132DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, -1 +; GFX1132DAGISEL-NEXT: s_ashr_i32 s6, s3, 31 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s7, s4, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s5, s5, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s4, s6 +; GFX1132DAGISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1132DAGISEL-NEXT: s_mul_i32 s4, s4, s3 +; GFX1132DAGISEL-NEXT: s_add_u32 s5, s5, s6 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s6, -1 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_ashr_i32 s7, s6, 31 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s9, s2, s6 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1132GISEL-NEXT: s_mul_i32 s7, s2, s7 +; GFX1132GISEL-NEXT: s_add_u32 s3, s9, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1132GISEL-NEXT: s_add_u32 s7, s3, s7 +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, -1 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_ashr_i32 s6, s3, 31 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_hi_u32 s7, s4, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s5, s5, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s8, s4, s6 +; GFX1132GISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s4, s3 +; GFX1132GISEL-NEXT: s_add_u32 s7, s5, s8 +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.sub.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.sub.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: ; GFX10DAGISEL: {{.*}} ; GFX10GISEL: {{.*}} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umax.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umax.ll index ba7a816184cd8..a4a5b01a873b7 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umax.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umax.ll @@ -947,3 +947,673 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: uniform_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: uniform_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umax.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v1, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: const_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v1, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: const_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v1, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, 0x7b +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v1, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132GISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umax.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: poison_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: poison_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_endpgm +; +; GFX11DAGISEL-LABEL: poison_value_i64: +; GFX11DAGISEL: ; %bb.0: ; %entry +; GFX11DAGISEL-NEXT: s_endpgm +; +; GFX11GISEL-LABEL: poison_value_i64: +; GFX11GISEL: ; %bb.0: ; %entry +; GFX11GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umax.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: divergent_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: divergent_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umax.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: ; %bb.1: ; %if +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: ; %bb.1: ; %if +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s4, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s4, exec_lo, s4 +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s6, exec_lo, s6 +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.umax.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.umax.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umin.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umin.ll index 3eaa89c957474..29a78855d6629 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umin.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umin.ll @@ -947,3 +947,865 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: uniform_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: uniform_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umin.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v1, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: const_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10DAGISEL-NEXT: global_store_dwordx2 v1, v[0:1], s[0:1] +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: const_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX10GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX10GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX10GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX10GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX10GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX10GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v1, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, 0x7b +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v1, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0x7b +; GFX1132GISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v2, 0 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umin.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_endpgm +; +; GFX10DAGISEL-LABEL: poison_value_i64: +; GFX10DAGISEL: ; %bb.0: ; %entry +; GFX10DAGISEL-NEXT: s_endpgm +; +; GFX10GISEL-LABEL: poison_value_i64: +; GFX10GISEL: ; %bb.0: ; %entry +; GFX10GISEL-NEXT: s_endpgm +; +; GFX11DAGISEL-LABEL: poison_value_i64: +; GFX11DAGISEL: ; %bb.0: ; %entry +; GFX11DAGISEL-NEXT: s_endpgm +; +; GFX11GISEL-LABEL: poison_value_i64: +; GFX11GISEL: ; %bb.0: ; %entry +; GFX11GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umin.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX8DAGISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[8:9], v[4:5] +; GFX8DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX8DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b32 s4, -1 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_mov_b32 s5, s4 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX8GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX8GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX8GISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[8:9], v[4:5] +; GFX8GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX8GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX9DAGISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[8:9], v[4:5] +; GFX9DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX9DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b32 s4, -1 +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_mov_b32 s5, s4 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX9GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX9GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX9GISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[8:9], v[4:5] +; GFX9GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX9GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[8:9], v[4:5] +; GFX1064DAGISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX1064DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b32 s4, -1 +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s12, s[6:7] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s12 +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v3, s12 +; GFX1064GISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[8:9], v[4:5] +; GFX1064GISEL-NEXT: s_and_b64 s[10:11], vcc, s[6:7] +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s12 +; GFX1064GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, -1 +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u64_e32 vcc_lo, s[8:9], v[4:5] +; GFX1032DAGISEL-NEXT: s_and_b32 s10, vcc_lo, s6 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s4, -1 +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v4, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v5, s5 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: v_cmp_lt_u64_e32 vcc_lo, s[8:9], v[4:5] +; GFX1032GISEL-NEXT: s_and_b32 s10, vcc_lo, s6 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_cselect_b64 s[4:5], s[8:9], s[4:5] +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, -1 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s8, s[2:3] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v4, s0 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v5, s1 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s8 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v3, s8 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164DAGISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[4:5], v[4:5] +; GFX1164DAGISEL-NEXT: s_and_b64 s[6:7], vcc, s[2:3] +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s8 +; GFX1164DAGISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b32 s0, -1 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s8, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v4, s0 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v5, s1 +; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s8 +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v3, s8 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmp_lt_u64_e32 vcc, s[4:5], v[4:5] +; GFX1164GISEL-NEXT: s_and_b64 s[6:7], vcc, s[2:3] +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s8 +; GFX1164GISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, -1 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132DAGISEL-NEXT: v_cmp_lt_u64_e32 vcc_lo, s[4:5], v[4:5] +; GFX1132DAGISEL-NEXT: s_and_b32 s6, vcc_lo, s2 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s0, -1 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v4, s0 :: v_dual_mov_b32 v5, s1 +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmp_lt_u64_e32 vcc_lo, s[4:5], v[4:5] +; GFX1132GISEL-NEXT: s_and_b32 s6, vcc_lo, s2 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_cselect_b64 s[0:1], s[4:5], s[0:1] +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.umin.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: ; %bb.1: ; %if +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: ; %bb.1: ; %if +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[4:5], s[4:5] +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s4, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s4, exec_lo, s4 +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b64 s[6:7], s[6:7] +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[6:7], s[6:7] +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s6, exec_lo, s6 +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %if +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.2: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s6 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[2:3] +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b64 s[6:7], s[4:5] +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.umin.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.umin.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.xor.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.xor.ll index 5b21d5c3aaeb6..b96954d030fef 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.xor.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.xor.ll @@ -1279,6 +1279,1419 @@ endif: store i32 %combine, ptr addrspace(1) %out ret void } + +define amdgpu_kernel void @uniform_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: uniform_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX8DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: s_mul_i32 s0, s2, s4 +; GFX8DAGISEL-NEXT: s_mul_i32 s1, s3, s4 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s4 +; GFX8DAGISEL-NEXT: s_add_u32 s1, s2, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: uniform_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX8GISEL-NEXT: s_and_b32 s5, s4, 1 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX8GISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: uniform_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9DAGISEL-NEXT: s_and_b32 s5, s4, 1 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: uniform_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9GISEL-NEXT: s_and_b32 s5, s4, 1 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9GISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: uniform_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: uniform_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064GISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1064GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1064GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: uniform_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: uniform_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032GISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1032GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: uniform_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: uniform_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1164GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1164GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1164GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: uniform_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1132DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: uniform_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1132GISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, s4 +; GFX1132GISEL-NEXT: s_add_u32 s3, s5, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.xor.i64(i64 %in, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @const_value_i64(ptr addrspace(1) %out) { +; GFX8DAGISEL-LABEL: const_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8DAGISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX8DAGISEL-NEXT: s_mul_i32 s4, s3, 0 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: const_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8GISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX8GISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX8GISEL-NEXT: s_mul_i32 s4, s3, 0 +; GFX8GISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s3 +; GFX8GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: const_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9DAGISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX9DAGISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s3, 0 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: const_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9GISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX9GISEL-NEXT: s_mul_i32 s2, s3, 0x7b +; GFX9GISEL-NEXT: s_mul_i32 s4, s3, 0 +; GFX9GISEL-NEXT: s_mul_hi_u32 s3, 0x7b, s3 +; GFX9GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: const_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1064DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: const_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1064GISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1064GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1064GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1064GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: const_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1032DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: const_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1032GISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1032GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1032GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: const_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1164DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: const_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1164GISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1164GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1164GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1164GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: const_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1132DAGISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: const_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1132GISEL-NEXT: s_mul_i32 s3, s2, 0 +; GFX1132GISEL-NEXT: s_mul_hi_u32 s4, 0x7b, s2 +; GFX1132GISEL-NEXT: s_mulk_i32 s2, 0x7b +; GFX1132GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.xor.i64(i64 123, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @poison_value_i64(ptr addrspace(1) %out, i64 %in) { +; GFX8DAGISEL-LABEL: poison_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8DAGISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX8DAGISEL-NEXT: s_mul_i32 s4, s1, s3 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s3, s0, s3 +; GFX8DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s3 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: poison_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX8GISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX8GISEL-NEXT: s_mul_i32 s4, s1, s3 +; GFX8GISEL-NEXT: s_mul_hi_u32 s3, s0, s3 +; GFX8GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: poison_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9DAGISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s1, s3 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s3, s0, s3 +; GFX9DAGISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: poison_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX9GISEL-NEXT: s_and_b32 s3, s2, 1 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s2, s0, s3 +; GFX9GISEL-NEXT: s_mul_i32 s4, s1, s3 +; GFX9GISEL-NEXT: s_mul_hi_u32 s3, s0, s3 +; GFX9GISEL-NEXT: s_add_u32 s3, s3, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: poison_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1064DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: poison_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1064GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1064GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1064GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1064GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: poison_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1032DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: poison_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24 +; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1032GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1032GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1032GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: poison_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1164DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: poison_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1164GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1164GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1164GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s3 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: poison_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132DAGISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1132DAGISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s3 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: poison_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_and_b32 s2, s2, 1 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_i32 s3, s1, s2 +; GFX1132GISEL-NEXT: s_mul_hi_u32 s4, s0, s2 +; GFX1132GISEL-NEXT: s_mul_i32 s2, s0, s2 +; GFX1132GISEL-NEXT: s_add_u32 s3, s4, s3 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.xor.i64(i64 poison, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) { +; GFX8DAGISEL-LABEL: divergent_value_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX8DAGISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8DAGISEL-NEXT: ; %bb.2: +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX8GISEL-LABEL: divergent_value_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX8GISEL-NEXT: s_mov_b32 s4, 0 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_mov_b32 s5, s4 +; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX8GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX8GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX8GISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX8GISEL-NEXT: ; %bb.2: +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[0:1], v[2:3] +; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX8GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9DAGISEL-LABEL: divergent_value_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX9DAGISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9DAGISEL-NEXT: ; %bb.2: +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX9GISEL-LABEL: divergent_value_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX9GISEL-NEXT: s_mov_b32 s4, 0 +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_mov_b32 s5, s4 +; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX9GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX9GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX9GISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX9GISEL-NEXT: ; %bb.2: +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX9GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) +; GFX9GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064DAGISEL-LABEL: divergent_value_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064DAGISEL-NEXT: ; %bb.2: +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1064GISEL-LABEL: divergent_value_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1064GISEL-NEXT: s_ff1_i32_b64 s10, s[6:7] +; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s10 +; GFX1064GISEL-NEXT: v_readlane_b32 s9, v3, s10 +; GFX1064GISEL-NEXT: s_bitset0_b64 s[6:7], s10 +; GFX1064GISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0 +; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1064GISEL-NEXT: ; %bb.2: +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1064GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032DAGISEL-LABEL: divergent_value_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032DAGISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032DAGISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032DAGISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032DAGISEL-NEXT: ; %bb.2: +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1032GISEL-LABEL: divergent_value_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mov_b32 s4, 0 +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_mov_b32 s5, s4 +; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6 +; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7 +; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7 +; GFX1032GISEL-NEXT: s_bitset0_b32 s6, s7 +; GFX1032GISEL-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9] +; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0 +; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1032GISEL-NEXT: ; %bb.2: +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5 +; GFX1032GISEL-NEXT: global_store_dwordx2 v[0:1], v[2:3], off +; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164DAGISEL-LABEL: divergent_value_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s6, s[2:3] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s6 +; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v3, s6 +; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s6 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[0:1], s[0:1], s[4:5] +; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164DAGISEL-NEXT: ; %bb.2: +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1164GISEL-LABEL: divergent_value_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec +; GFX1164GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1164GISEL-NEXT: s_ctz_i32_b64 s6, s[2:3] +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s6 +; GFX1164GISEL-NEXT: v_readlane_b32 s5, v3, s6 +; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s6 +; GFX1164GISEL-NEXT: s_xor_b64 s[0:1], s[0:1], s[4:5] +; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0 +; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1164GISEL-NEXT: ; %bb.2: +; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX1164GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132DAGISEL-LABEL: divergent_value_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132DAGISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132DAGISEL-NEXT: s_xor_b64 s[0:1], s[0:1], s[4:5] +; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132DAGISEL-NEXT: ; %bb.2: +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31] +; +; GFX1132GISEL-LABEL: divergent_value_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mov_b32 s0, 0 +; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo +; GFX1132GISEL-NEXT: s_mov_b32 s1, s0 +; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1 +; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2 +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3 +; GFX1132GISEL-NEXT: v_readlane_b32 s5, v3, s3 +; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3 +; GFX1132GISEL-NEXT: s_xor_b64 s[0:1], s[0:1], s[4:5] +; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0 +; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1 +; GFX1132GISEL-NEXT: ; %bb.2: +; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0 +; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off +; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31] +entry: + %result = call i64 @llvm.amdgcn.wave.reduce.xor.i64(i64 %id.x, i32 1) + store i64 %result, ptr addrspace(1) %out + ret void +} + +define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64 %in2) { +; GFX8DAGISEL-LABEL: divergent_cfg_i64: +; GFX8DAGISEL: ; %bb.0: ; %entry +; GFX8DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8DAGISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8DAGISEL-NEXT: ; %bb.1: ; %else +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8DAGISEL-NEXT: s_and_b32 s7, s6, 1 +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX8DAGISEL-NEXT: s_add_u32 s7, s2, s3 +; GFX8DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX8DAGISEL-NEXT: ; %bb.3: ; %if +; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8DAGISEL-NEXT: s_and_b32 s7, s6, 1 +; GFX8DAGISEL-NEXT: s_mul_i32 s6, s4, s7 +; GFX8DAGISEL-NEXT: s_mul_i32 s5, s5, s7 +; GFX8DAGISEL-NEXT: s_mul_hi_u32 s4, s4, s7 +; GFX8DAGISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8DAGISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8DAGISEL-NEXT: s_endpgm +; +; GFX8GISEL-LABEL: divergent_cfg_i64: +; GFX8GISEL: ; %bb.0: ; %entry +; GFX8GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX8GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX8GISEL-NEXT: ; %bb.1: ; %else +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8GISEL-NEXT: s_and_b32 s7, s6, 1 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX8GISEL-NEXT: s_add_u32 s7, s2, s3 +; GFX8GISEL-NEXT: .LBB9_2: ; %Flow +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX8GISEL-NEXT: ; %bb.3: ; %if +; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34 +; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX8GISEL-NEXT: s_and_b32 s7, s6, 1 +; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX8GISEL-NEXT: s_mul_i32 s6, s4, s7 +; GFX8GISEL-NEXT: s_mul_i32 s5, s5, s7 +; GFX8GISEL-NEXT: s_mul_hi_u32 s4, s4, s7 +; GFX8GISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX8GISEL-NEXT: .LBB9_4: ; %endif +; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1 +; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0 +; GFX8GISEL-NEXT: flat_store_dwordx2 v[2:3], v[0:1] +; GFX8GISEL-NEXT: s_endpgm +; +; GFX9DAGISEL-LABEL: divergent_cfg_i64: +; GFX9DAGISEL: ; %bb.0: ; %entry +; GFX9DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX9DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9DAGISEL-NEXT: ; %bb.1: ; %else +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9DAGISEL-NEXT: s_and_b32 s5, s4, 1 +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s2, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s3 +; GFX9DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX9DAGISEL-NEXT: ; %bb.3: ; %if +; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9DAGISEL-NEXT: s_and_b32 s5, s4, 1 +; GFX9DAGISEL-NEXT: s_mul_i32 s4, s6, s5 +; GFX9DAGISEL-NEXT: s_mul_i32 s7, s7, s5 +; GFX9DAGISEL-NEXT: s_mul_hi_u32 s5, s6, s5 +; GFX9DAGISEL-NEXT: s_add_u32 s5, s5, s7 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX9DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9DAGISEL-NEXT: s_endpgm +; +; GFX9GISEL-LABEL: divergent_cfg_i64: +; GFX9GISEL: ; %bb.0: ; %entry +; GFX9GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX9GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX9GISEL-NEXT: ; %bb.1: ; %else +; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX9GISEL-NEXT: s_and_b32 s7, s6, 1 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s6, s2, s7 +; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s7 +; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s7 +; GFX9GISEL-NEXT: s_add_u32 s7, s2, s3 +; GFX9GISEL-NEXT: .LBB9_2: ; %Flow +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX9GISEL-NEXT: ; %bb.3: ; %if +; GFX9GISEL-NEXT: s_load_dwordx2 s[8:9], s[4:5], 0x34 +; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX9GISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX9GISEL-NEXT: s_mul_i32 s6, s8, s4 +; GFX9GISEL-NEXT: s_mul_i32 s5, s9, s4 +; GFX9GISEL-NEXT: s_mul_hi_u32 s4, s8, s4 +; GFX9GISEL-NEXT: s_add_u32 s7, s4, s5 +; GFX9GISEL-NEXT: .LBB9_4: ; %endif +; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX9GISEL-NEXT: s_endpgm +; +; GFX1064DAGISEL-LABEL: divergent_cfg_i64: +; GFX1064DAGISEL: ; %bb.0: ; %entry +; GFX1064DAGISEL-NEXT: s_clause 0x1 +; GFX1064DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc, 15, v0 +; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9 +; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5] +; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1064DAGISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9] +; GFX1064DAGISEL-NEXT: s_and_b32 s8, s8, 1 +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s8 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s9, s2, s8 +; GFX1064DAGISEL-NEXT: s_mul_i32 s8, s2, s8 +; GFX1064DAGISEL-NEXT: s_add_u32 s9, s9, s3 +; GFX1064DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[4:5] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s8 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s9 +; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX1064DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1064DAGISEL-NEXT: s_mul_i32 s5, s7, s4 +; GFX1064DAGISEL-NEXT: s_mul_hi_u32 s7, s6, s4 +; GFX1064DAGISEL-NEXT: s_mul_i32 s4, s6, s4 +; GFX1064DAGISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1064DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064DAGISEL-NEXT: s_endpgm +; +; GFX1064GISEL-LABEL: divergent_cfg_i64: +; GFX1064GISEL: ; %bb.0: ; %entry +; GFX1064GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1064GISEL-NEXT: v_cmp_le_u32_e32 vcc, 16, v0 +; GFX1064GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc +; GFX1064GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1064GISEL-NEXT: ; %bb.1: ; %else +; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1064GISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1064GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1064GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1064GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_andn2_saveexec_b64 s[2:3], s[8:9] +; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1064GISEL-NEXT: ; %bb.3: ; %if +; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec +; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5] +; GFX1064GISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1064GISEL-NEXT: s_mul_i32 s5, s7, s4 +; GFX1064GISEL-NEXT: s_mul_hi_u32 s7, s6, s4 +; GFX1064GISEL-NEXT: s_mul_i32 s6, s6, s4 +; GFX1064GISEL-NEXT: s_add_u32 s7, s7, s5 +; GFX1064GISEL-NEXT: .LBB9_4: ; %endif +; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1064GISEL-NEXT: s_endpgm +; +; GFX1032DAGISEL-LABEL: divergent_cfg_i64: +; GFX1032DAGISEL: ; %bb.0: ; %entry +; GFX1032DAGISEL-NEXT: s_clause 0x1 +; GFX1032DAGISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032DAGISEL-NEXT: v_cmp_lt_u32_e32 vcc_lo, 15, v0 +; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5 +; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4 +; GFX1032DAGISEL-NEXT: s_and_b32 s4, s4, 1 +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s5, s2, s4 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s2, s4 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s5, s3 +; GFX1032DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s2, s8 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2 +; GFX1032DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1032DAGISEL-NEXT: s_and_b32 s3, s3, 1 +; GFX1032DAGISEL-NEXT: s_mul_i32 s5, s7, s3 +; GFX1032DAGISEL-NEXT: s_mul_hi_u32 s7, s6, s3 +; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s6, s3 +; GFX1032DAGISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1032DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032DAGISEL-NEXT: s_endpgm +; +; GFX1032GISEL-LABEL: divergent_cfg_i64: +; GFX1032GISEL: ; %bb.0: ; %entry +; GFX1032GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24 +; GFX1032GISEL-NEXT: v_cmp_le_u32_e32 vcc_lo, 16, v0 +; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo +; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1032GISEL-NEXT: ; %bb.1: ; %else +; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1032GISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1032GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_andn2_saveexec_b32 s2, s8 +; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1032GISEL-NEXT: ; %bb.3: ; %if +; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34 +; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1032GISEL-NEXT: s_and_b32 s3, s3, 1 +; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1032GISEL-NEXT: s_mul_i32 s4, s7, s3 +; GFX1032GISEL-NEXT: s_mul_hi_u32 s5, s6, s3 +; GFX1032GISEL-NEXT: s_mul_i32 s6, s6, s3 +; GFX1032GISEL-NEXT: s_add_u32 s7, s5, s4 +; GFX1032GISEL-NEXT: .LBB9_4: ; %endif +; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1] +; GFX1032GISEL-NEXT: s_endpgm +; +; GFX1164DAGISEL-LABEL: divergent_cfg_i64: +; GFX1164DAGISEL: ; %bb.0: ; %entry +; GFX1164DAGISEL-NEXT: s_clause 0x1 +; GFX1164DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9 +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7] +; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1164DAGISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9] +; GFX1164DAGISEL-NEXT: s_and_b32 s8, s8, 1 +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s8 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s9, s2, s8 +; GFX1164DAGISEL-NEXT: s_mul_i32 s8, s2, s8 +; GFX1164DAGISEL-NEXT: s_add_u32 s9, s9, s3 +; GFX1164DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s8 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s9 +; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3] +; GFX1164DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164DAGISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1164DAGISEL-NEXT: s_mul_i32 s5, s5, s6 +; GFX1164DAGISEL-NEXT: s_mul_hi_u32 s7, s4, s6 +; GFX1164DAGISEL-NEXT: s_mul_i32 s4, s4, s6 +; GFX1164DAGISEL-NEXT: s_add_u32 s5, s7, s5 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4 +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s5 +; GFX1164DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164DAGISEL-NEXT: s_endpgm +; +; GFX1164GISEL-LABEL: divergent_cfg_i64: +; GFX1164GISEL: ; %bb.0: ; %entry +; GFX1164GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec +; GFX1164GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1164GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1164GISEL-NEXT: ; %bb.1: ; %else +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164GISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1164GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1164GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_and_not1_saveexec_b64 s[2:3], s[8:9] +; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1164GISEL-NEXT: ; %bb.3: ; %if +; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec +; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7] +; GFX1164GISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1164GISEL-NEXT: s_mul_i32 s5, s5, s6 +; GFX1164GISEL-NEXT: s_mul_hi_u32 s7, s4, s6 +; GFX1164GISEL-NEXT: s_mul_i32 s6, s4, s6 +; GFX1164GISEL-NEXT: s_add_u32 s7, s7, s5 +; GFX1164GISEL-NEXT: .LBB9_4: ; %endif +; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3] +; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s6 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s7 +; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1164GISEL-NEXT: s_endpgm +; +; GFX1132DAGISEL-LABEL: divergent_cfg_i64: +; GFX1132DAGISEL: ; %bb.0: ; %entry +; GFX1132DAGISEL-NEXT: s_clause 0x1 +; GFX1132DAGISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132DAGISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132DAGISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0 +; GFX1132DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else +; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1132DAGISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1132DAGISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1132DAGISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s2, s8 +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2 +; GFX1132DAGISEL-NEXT: ; %bb.3: ; %if +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: s_and_b32 s3, s3, 1 +; GFX1132DAGISEL-NEXT: s_mul_i32 s5, s5, s3 +; GFX1132DAGISEL-NEXT: s_mul_hi_u32 s6, s4, s3 +; GFX1132DAGISEL-NEXT: s_mul_i32 s4, s4, s3 +; GFX1132DAGISEL-NEXT: s_add_u32 s5, s6, s5 +; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) +; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5 +; GFX1132DAGISEL-NEXT: ; %bb.4: ; %endif +; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132DAGISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132DAGISEL-NEXT: s_endpgm +; +; GFX1132GISEL-LABEL: divergent_cfg_i64: +; GFX1132GISEL: ; %bb.0: ; %entry +; GFX1132GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24 +; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0 +; GFX1132GISEL-NEXT: s_mov_b32 s8, exec_lo +; GFX1132GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7 +; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) +; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0 +; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2 +; GFX1132GISEL-NEXT: ; %bb.1: ; %else +; GFX1132GISEL-NEXT: s_mov_b32 s6, exec_lo +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s6, s6 +; GFX1132GISEL-NEXT: s_and_b32 s6, s6, 1 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s6 +; GFX1132GISEL-NEXT: s_mul_hi_u32 s7, s2, s6 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s2, s6 +; GFX1132GISEL-NEXT: s_add_u32 s7, s7, s3 +; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_and_not1_saveexec_b32 s2, s8 +; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4 +; GFX1132GISEL-NEXT: ; %bb.3: ; %if +; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34 +; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo +; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) +; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s3, s3 +; GFX1132GISEL-NEXT: s_and_b32 s3, s3, 1 +; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0) +; GFX1132GISEL-NEXT: s_mul_i32 s5, s5, s3 +; GFX1132GISEL-NEXT: s_mul_hi_u32 s7, s4, s3 +; GFX1132GISEL-NEXT: s_mul_i32 s6, s4, s3 +; GFX1132GISEL-NEXT: s_add_u32 s7, s7, s5 +; GFX1132GISEL-NEXT: .LBB9_4: ; %endif +; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2 +; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7 +; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0 +; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1] +; GFX1132GISEL-NEXT: s_endpgm +entry: + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %d_cmp = icmp ult i32 %tid, 16 + br i1 %d_cmp, label %if, label %else + +if: + %reducedValTid = call i64 @llvm.amdgcn.wave.reduce.xor.i64(i64 %in2, i32 1) + br label %endif + +else: + %reducedValIn = call i64 @llvm.amdgcn.wave.reduce.xor.i64(i64 %in, i32 1) + br label %endif + +endif: + %combine = phi i64 [%reducedValTid, %if], [%reducedValIn, %else] + store i64 %combine, ptr addrspace(1) %out + ret void +} ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line: ; GFX10DAGISEL: {{.*}} ; GFX10GISEL: {{.*}}