diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2017-02-21 23:35:48 +0000 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2017-02-21 23:35:48 +0000 |
| commit | 2fdf2a1a189ce2d0462eb1902a88f5210b963063 (patch) | |
| tree | cf28f9b484ce67fa9098bb72532749abe3f54aa0 /llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp | |
| parent | e6e206d4b48146b194b04f43a3dec4d32d568019 (diff) | |
| download | bcm5719-llvm-2fdf2a1a189ce2d0462eb1902a88f5210b963063.tar.gz bcm5719-llvm-2fdf2a1a189ce2d0462eb1902a88f5210b963063.zip | |
AMDGPU: Redefine clamp node as clamp 0.0-1.0
Change implementation to use max instead of add.
min/max/med3 do not flush denormals regardless of the mode,
so it is OK to use it whether or not they are enabled.
Also allow using clamp with f16, and use knowledge
of dx10_clamp.
llvm-svn: 295788
Diffstat (limited to 'llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp')
| -rw-r--r-- | llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp b/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp index 32b04fe7a1e..a719931eff0 100644 --- a/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp +++ b/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp @@ -42,7 +42,7 @@ AMDGPUSubtarget::initializeSubtargetDependencies(const Triple &TT, // for SI has the unhelpful behavior that it unsets everything else if you // disable it. - SmallString<256> FullFS("+promote-alloca,+fp64-fp16-denormals,+load-store-opt,"); + SmallString<256> FullFS("+promote-alloca,+fp64-fp16-denormals,+dx10-clamp,+load-store-opt,"); if (isAmdHsaOS()) // Turn on FlatForGlobal for HSA. FullFS += "+flat-for-global,+unaligned-buffer-access,+trap-handler,"; @@ -89,6 +89,7 @@ AMDGPUSubtarget::AMDGPUSubtarget(const Triple &TT, StringRef GPU, StringRef FS, FP32Denormals(false), FP64FP16Denormals(false), FPExceptions(false), + DX10Clamp(false), FlatForGlobal(false), UnalignedScratchAccess(false), UnalignedBufferAccess(false), |
