diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2016-12-22 03:05:37 +0000 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2016-12-22 03:05:37 +0000 |
| commit | ce84130f8562c8c990362502f03d04187a0be581 (patch) | |
| tree | 8bf3fdd9e71571683d331712f2337702e36305dd /llvm/lib/Target/AMDGPU | |
| parent | 4e55c1ec116de59ea1a0225974d616ac9236c43f (diff) | |
| download | bcm5719-llvm-ce84130f8562c8c990362502f03d04187a0be581.tar.gz bcm5719-llvm-ce84130f8562c8c990362502f03d04187a0be581.zip | |
AMDGPU: Implement f16 fcanonicalize
llvm-svn: 290300
Diffstat (limited to 'llvm/lib/Target/AMDGPU')
| -rw-r--r-- | llvm/lib/Target/AMDGPU/AMDGPUInstructions.td | 1 | ||||
| -rw-r--r-- | llvm/lib/Target/AMDGPU/SIISelLowering.cpp | 3 | ||||
| -rw-r--r-- | llvm/lib/Target/AMDGPU/SIInstructions.td | 5 |
3 files changed, 9 insertions, 0 deletions
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUInstructions.td b/llvm/lib/Target/AMDGPU/AMDGPUInstructions.td index a7a995156e2..c49866da5d3 100644 --- a/llvm/lib/Target/AMDGPU/AMDGPUInstructions.td +++ b/llvm/lib/Target/AMDGPU/AMDGPUInstructions.td @@ -391,6 +391,7 @@ int TWO_PI = 0x40c90fdb; int PI = 0x40490fdb; int TWO_PI_INV = 0x3e22f983; int FP_UINT_MAX_PLUS_1 = 0x4f800000; // 1 << 32 in floating point encoding +int FP16_ONE = 0x3C00; int FP32_ONE = 0x3f800000; int FP32_NEG_ONE = 0xbf800000; int FP64_ONE = 0x3ff0000000000000; diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp index 0558c59782c..b9302582fa8 100644 --- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp +++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp @@ -3648,6 +3648,9 @@ SDValue SITargetLowering::performFCanonicalizeCombine( if (VT == MVT::f64 && !Subtarget->hasFP64Denormals()) return DAG.getConstantFP(0.0, SDLoc(N), VT); + + if (VT == MVT::f16 && !Subtarget->hasFP16Denormals()) + return DAG.getConstantFP(0.0, SDLoc(N), VT); } if (C.isNaN()) { diff --git a/llvm/lib/Target/AMDGPU/SIInstructions.td b/llvm/lib/Target/AMDGPU/SIInstructions.td index 54865956ae2..bc35c2edc8d 100644 --- a/llvm/lib/Target/AMDGPU/SIInstructions.td +++ b/llvm/lib/Target/AMDGPU/SIInstructions.td @@ -1022,6 +1022,11 @@ defm : BFMPatterns <i32, S_BFM_B32, S_MOV_B32>; def : BFEPattern <V_BFE_U32, S_MOV_B32>; def : Pat< + (fcanonicalize f16:$src), + (V_MUL_F16_e64 0, (i32 CONST.FP16_ONE), 0, $src, 0, 0) +>; + +def : Pat< (fcanonicalize f32:$src), (V_MUL_F32_e64 0, (i32 CONST.FP32_ONE), 0, $src, 0, 0) >; |
