From 76e4afb063c8fb0c87185d8d08bd701ca6b5a558 Mon Sep 17 00:00:00 2001 From: Changpeng Fang Date: Tue, 28 Oct 2025 17:11:54 -0700 Subject: [PATCH 1/2] [AMDGPU] Support bfloat comparison for ballot intrinsic We do not have native instructions for direct bfloat comparisons. However, we can expand bfloat to float, and do float coparison instead. TODO: handle bfloat comparison for ballot intrinsic on global isel path. Fixes: SWDEV-563403 --- llvm/lib/Target/AMDGPU/SIISelLowering.cpp | 10 +++++++-- .../CodeGen/AMDGPU/llvm.amdgcn.ballot.i32.ll | 21 +++++++++++++++++++ .../CodeGen/AMDGPU/llvm.amdgcn.ballot.i64.ll | 12 +++++++++++ 3 files changed, 41 insertions(+), 2 deletions(-) diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp index be4229155c983..2c90ec2088e45 100644 --- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp +++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp @@ -7035,9 +7035,15 @@ static SDValue lowerBALLOTIntrinsic(const SITargetLowering &TLI, SDNode *N, SDLoc SL(N); if (Src.getOpcode() == ISD::SETCC) { + SDValue Op0 = Src.getOperand(0); + SDValue Op1 = Src.getOperand(1); + // Need to expand bflat to float for comparison (setcc). + if (Op0.getValueType() == MVT::bf16) { + Op0 = DAG.getNode(ISD::FP_EXTEND, SL, MVT::f32, Op0); + Op1 = DAG.getNode(ISD::FP_EXTEND, SL, MVT::f32, Op1); + } // (ballot (ISD::SETCC ...)) -> (AMDGPUISD::SETCC ...) - return DAG.getNode(AMDGPUISD::SETCC, SL, VT, Src.getOperand(0), - Src.getOperand(1), Src.getOperand(2)); + return DAG.getNode(AMDGPUISD::SETCC, SL, VT, Op0, Op1, Src.getOperand(2)); } if (const ConstantSDNode *Arg = dyn_cast(Src)) { // (ballot 0) -> 0 diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i32.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i32.ll index e00e1f13b2b77..9940ea70d3467 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i32.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i32.ll @@ -591,3 +591,24 @@ exit: store i32 %ballot, ptr addrspace(1) %out ret void } + +define amdgpu_cs i32 @compare_bfloats(bfloat %x, bfloat %y) { +; GFX10-LABEL: compare_bfloats: +; GFX10: ; %bb.0: +; GFX10-NEXT: v_lshlrev_b32_e32 v1, 16, v1 +; GFX10-NEXT: v_lshlrev_b32_e32 v0, 16, v0 +; GFX10-NEXT: v_cmp_gt_f32_e64 s0, v0, v1 +; GFX10-NEXT: ; return to shader part epilog +; +; GFX11-LABEL: compare_bfloats: +; GFX11: ; %bb.0: +; GFX11-NEXT: v_mov_b16_e32 v2.l, 0 +; GFX11-NEXT: v_mov_b16_e32 v2.h, v1.l +; GFX11-NEXT: v_mov_b16_e32 v1.h, v0.l +; GFX11-NEXT: v_mov_b16_e32 v1.l, v2.l +; GFX11-NEXT: v_cmp_gt_f32_e64 s0, v1, v2 +; GFX11-NEXT: ; return to shader part epilog + %cmp = fcmp ogt bfloat %x, %y + %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 %cmp) + ret i32 %ballot +} diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i64.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i64.ll index b4adf7f641550..1720a62eb6367 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i64.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i64.ll @@ -557,3 +557,15 @@ exit: store i64 %ballot, ptr addrspace(1) %out ret void } + +define amdgpu_cs i64 @compare_bfloats(bfloat %x, bfloat %y) { +; CHECK-LABEL: compare_bfloats: +; CHECK: ; %bb.0: +; CHECK-NEXT: v_lshlrev_b32_e32 v1, 16, v1 +; CHECK-NEXT: v_lshlrev_b32_e32 v0, 16, v0 +; CHECK-NEXT: v_cmp_gt_f32_e64 s[0:1], v0, v1 +; CHECK-NEXT: ; return to shader part epilog + %cmp = fcmp ogt bfloat %x, %y + %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %cmp) + ret i64 %ballot +} From a84e5e869d71918e5d54ce398e4fa6d292297e18 Mon Sep 17 00:00:00 2001 From: Changpeng Fang Date: Tue, 28 Oct 2025 21:11:41 -0700 Subject: [PATCH 2/2] Fix a typo --- llvm/lib/Target/AMDGPU/SIISelLowering.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp index 2c90ec2088e45..adcc5e5ff1044 100644 --- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp +++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp @@ -7037,7 +7037,7 @@ static SDValue lowerBALLOTIntrinsic(const SITargetLowering &TLI, SDNode *N, if (Src.getOpcode() == ISD::SETCC) { SDValue Op0 = Src.getOperand(0); SDValue Op1 = Src.getOperand(1); - // Need to expand bflat to float for comparison (setcc). + // Need to expand bfloat to float for comparison (setcc). if (Op0.getValueType() == MVT::bf16) { Op0 = DAG.getNode(ISD::FP_EXTEND, SL, MVT::f32, Op0); Op1 = DAG.getNode(ISD::FP_EXTEND, SL, MVT::f32, Op1);