Skip to content

[HLSL] Add WaveReadLaneAt unsigned integer overloads #133520

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
spall merged 1 commit into from
Mar 28, 2025
Merged

[HLSL] Add WaveReadLaneAt unsigned integer overloads #133520

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
87 changes: 59 additions & 28 deletions clang/lib/Headers/hlsl/hlsl_alias_intrinsics.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2358,77 +2358,108 @@ __attribute__((convergent)) bool WaveIsFirstLane();
// the specified wave.

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) bool WaveReadLaneAt(bool, int32_t);
__attribute__((convergent)) bool WaveReadLaneAt(bool, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) bool2 WaveReadLaneAt(bool2, int32_t);
__attribute__((convergent)) bool2 WaveReadLaneAt(bool2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) bool3 WaveReadLaneAt(bool3, int32_t);
__attribute__((convergent)) bool3 WaveReadLaneAt(bool3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) bool4 WaveReadLaneAt(bool4, int32_t);
__attribute__((convergent)) bool4 WaveReadLaneAt(bool4, uint32_t);

#ifdef __HLSL_ENABLE_16_BIT
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int16_t WaveReadLaneAt(int16_t, int32_t);
__attribute__((convergent)) int16_t WaveReadLaneAt(int16_t, uint32_t);
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int16_t2 WaveReadLaneAt(int16_t2, int32_t);
__attribute__((convergent)) int16_t2 WaveReadLaneAt(int16_t2, uint32_t);
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int16_t3 WaveReadLaneAt(int16_t3, int32_t);
__attribute__((convergent)) int16_t3 WaveReadLaneAt(int16_t3, uint32_t);
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int16_t4 WaveReadLaneAt(int16_t4, int32_t);
__attribute__((convergent)) int16_t4 WaveReadLaneAt(int16_t4, uint32_t);

_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint16_t WaveReadLaneAt(uint16_t, uint32_t);
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint16_t2 WaveReadLaneAt(uint16_t2, uint32_t);
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint16_t3 WaveReadLaneAt(uint16_t3, uint32_t);
_HLSL_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint16_t4 WaveReadLaneAt(uint16_t4, uint32_t);
#endif

_HLSL_16BIT_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) half WaveReadLaneAt(half, int32_t);
__attribute__((convergent)) half WaveReadLaneAt(half, uint32_t);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) half2 WaveReadLaneAt(half2, int32_t);
__attribute__((convergent)) half2 WaveReadLaneAt(half2, uint32_t);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) half3 WaveReadLaneAt(half3, int32_t);
__attribute__((convergent)) half3 WaveReadLaneAt(half3, uint32_t);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.0)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) half4 WaveReadLaneAt(half4, int32_t);
__attribute__((convergent)) half4 WaveReadLaneAt(half4, uint32_t);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int WaveReadLaneAt(int, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int2 WaveReadLaneAt(int2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int3 WaveReadLaneAt(int3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int4 WaveReadLaneAt(int4, uint32_t);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint WaveReadLaneAt(uint, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint2 WaveReadLaneAt(uint2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint3 WaveReadLaneAt(uint3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) uint4 WaveReadLaneAt(uint4, uint32_t);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int WaveReadLaneAt(int, int32_t);
__attribute__((convergent)) float WaveReadLaneAt(float, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int2 WaveReadLaneAt(int2, int32_t);
__attribute__((convergent)) float2 WaveReadLaneAt(float2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int3 WaveReadLaneAt(int3, int32_t);
__attribute__((convergent)) float3 WaveReadLaneAt(float3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int4 WaveReadLaneAt(int4, int32_t);
__attribute__((convergent)) float4 WaveReadLaneAt(float4, uint32_t);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) float WaveReadLaneAt(float, int32_t);
__attribute__((convergent)) int64_t WaveReadLaneAt(int64_t, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) float2 WaveReadLaneAt(float2, int32_t);
__attribute__((convergent)) int64_t2 WaveReadLaneAt(int64_t2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) float3 WaveReadLaneAt(float3, int32_t);
__attribute__((convergent)) int64_t3 WaveReadLaneAt(int64_t3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) float4 WaveReadLaneAt(float4, int32_t);
__attribute__((convergent)) int64_t4 WaveReadLaneAt(int64_t4, uint32_t);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int64_t WaveReadLaneAt(int64_t, int32_t);
__attribute__((convergent)) uint64_t WaveReadLaneAt(uint64_t, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int64_t2 WaveReadLaneAt(int64_t2, int32_t);
__attribute__((convergent)) uint64_t2 WaveReadLaneAt(uint64_t2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int64_t3 WaveReadLaneAt(int64_t3, int32_t);
__attribute__((convergent)) uint64_t3 WaveReadLaneAt(uint64_t3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) int64_t4 WaveReadLaneAt(int64_t4, int32_t);
__attribute__((convergent)) uint64_t4 WaveReadLaneAt(uint64_t4, uint32_t);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) double WaveReadLaneAt(double, int32_t);
__attribute__((convergent)) double WaveReadLaneAt(double, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) double2 WaveReadLaneAt(double2, int32_t);
__attribute__((convergent)) double2 WaveReadLaneAt(double2, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) double3 WaveReadLaneAt(double3, int32_t);
__attribute__((convergent)) double3 WaveReadLaneAt(double3, uint32_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_wave_read_lane_at)
__attribute__((convergent)) double4 WaveReadLaneAt(double4, int32_t);
__attribute__((convergent)) double4 WaveReadLaneAt(double4, uint32_t);

//===----------------------------------------------------------------------===//
// WaveActiveMax builtins
Expand Down
39 changes: 39 additions & 0 deletions clang/test/CodeGenHLSL/builtins/WaveReadLaneAt.hlsl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,36 @@ int test_int(int expr, uint idx) {
// CHECK-DXIL: declare [[TY]] @llvm.dx.wave.readlane.i32([[TY]], i32) #[[#attr:]]
// CHECK-SPIRV: declare spir_func [[TY]] @llvm.spv.wave.readlane.i32([[TY]], i32) #[[#attr:]]

// CHECK-LABEL: test_uint
uint test_uint(uint expr, uint idx) {
// CHECK-SPIRV: %[[#entry_tok0:]] = call token @llvm.experimental.convergence.entry()
// CHECK-SPIRV: %[[RET:.*]] = call spir_func [[TY:.*]] @llvm.spv.wave.readlane.i32([[TY]] %[[#]], i32 %[[#]]) [ "convergencectrl"(token %[[#entry_tok0]]) ]
// CHECK-DXIL: %[[RET:.*]] = call [[TY:.*]] @llvm.dx.wave.readlane.i32([[TY]] %[[#]], i32 %[[#]])
// CHECK: ret [[TY]] %[[RET]]
return WaveReadLaneAt(expr, idx);
}

// CHECK-LABEL: test_int64_t
int64_t test_int64_t(int64_t expr, uint idx) {
// CHECK-SPIRV: %[[#entry_tok0:]] = call token @llvm.experimental.convergence.entry()
// CHECK-SPIRV: %[[RET:.*]] = call spir_func [[TY:.*]] @llvm.spv.wave.readlane.i64([[TY]] %[[#]], i32 %[[#]]) [ "convergencectrl"(token %[[#entry_tok0]]) ]
// CHECK-DXIL: %[[RET:.*]] = call [[TY:.*]] @llvm.dx.wave.readlane.i64([[TY]] %[[#]], i32 %[[#]])
// CHECK: ret [[TY]] %[[RET]]
return WaveReadLaneAt(expr, idx);
}

// CHECK-DXIL: declare [[TY]] @llvm.dx.wave.readlane.i64([[TY]], i32) #[[#attr:]]
// CHECK-SPIRV: declare spir_func [[TY]] @llvm.spv.wave.readlane.i64([[TY]], i32) #[[#attr:]]

// CHECK-LABEL: test_uint64_t
uint64_t test_uint64_t(uint64_t expr, uint idx) {
// CHECK-SPIRV: %[[#entry_tok0:]] = call token @llvm.experimental.convergence.entry()
// CHECK-SPIRV: %[[RET:.*]] = call spir_func [[TY:.*]] @llvm.spv.wave.readlane.i64([[TY]] %[[#]], i32 %[[#]]) [ "convergencectrl"(token %[[#entry_tok0]]) ]
// CHECK-DXIL: %[[RET:.*]] = call [[TY:.*]] @llvm.dx.wave.readlane.i64([[TY]] %[[#]], i32 %[[#]])
// CHECK: ret [[TY]] %[[RET]]
return WaveReadLaneAt(expr, idx);
}

#ifdef __HLSL_ENABLE_16_BIT
// CHECK-LABEL: test_int16
int16_t test_int16(int16_t expr, uint idx) {
Expand All @@ -31,6 +61,15 @@ int16_t test_int16(int16_t expr, uint idx) {

// CHECK-DXIL: declare [[TY]] @llvm.dx.wave.readlane.i16([[TY]], i32) #[[#attr:]]
// CHECK-SPIRV: declare spir_func [[TY]] @llvm.spv.wave.readlane.i16([[TY]], i32) #[[#attr:]]

// CHECK-LABEL: test_uint16
uint16_t test_uint16(uint16_t expr, uint idx) {
// CHECK-SPIRV: %[[#entry_tok1:]] = call token @llvm.experimental.convergence.entry()
// CHECK-SPIRV: %[[RET:.*]] = call spir_func [[TY:.*]] @llvm.spv.wave.readlane.i16([[TY]] %[[#]], i32 %[[#]]) [ "convergencectrl"(token %[[#entry_tok1]]) ]
// CHECK-DXIL: %[[RET:.*]] = call [[TY:.*]] @llvm.dx.wave.readlane.i16([[TY]] %[[#]], i32 %[[#]])
// CHECK: ret [[TY]] %[[RET]]
return WaveReadLaneAt(expr, idx);
}
#endif

// Test basic lowering to runtime function call with array and float values.
Expand Down