[PATCH] D158997: AMDGPU: Use exp2 and log2 intrinsics directly for f16/f32

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added reviewers: AMDGPU, rampitec, jmmartinez.
Herald added subscribers: foad, kerbowa, hiraditya, tpr, dstuttard, yaxunl, jvesely, kzhuravl.
Herald added a project: All.
arsenm requested review of this revision.
Herald added a subscriber: wdng.
Herald added a project: LLVM.

These codegen correctly but f64 doesn't. This prevents losing fast
math flags on the way to the underlying intrinsic.


https://reviews.llvm.org/D158997

Files:
  llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp
  llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-pow-codegen.ll
  llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-pow.ll
  llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-pown.ll
  llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-powr.ll
  llvm/test/CodeGen/AMDGPU/simplify-libcalls.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D158997.553950.patch
Type: text/x-patch
Size: 107022 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230828/39f29445/attachment-0001.bin>