[all-commits] [llvm/llvm-project] f337a7: AMDGPU: Replace rounding libcalls with intrinsics

[Matt Arsenault via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: f337a77c99aa31b37c60dd2ecbd96f8317426fad
      https://github.com/llvm/llvm-project/commit/f337a77c99aa31b37c60dd2ecbd96f8317426fad
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-08-14 (Mon, 14 Aug 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp
    M llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-ceil.ll
    M llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-floor.ll
    M llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-rint.ll
    M llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-round.ll
    M llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-trunc.ll

  Log Message:
  -----------
  AMDGPU: Replace rounding libcalls with intrinsics


  Commit: 0eabe65bfbb8eed5d0ffb3718fed3c9bbe058379
      https://github.com/llvm/llvm-project/commit/0eabe65bfbb8eed5d0ffb3718fed3c9bbe058379
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-08-14 (Mon, 14 Aug 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp
    M llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-ldexp.ll

  Log Message:
  -----------
  AMDGPU: Replace ldexp libcalls with intrinsic


  Commit: 416f6af9765cba2f9c45901b07e8d793f482f26d
      https://github.com/llvm/llvm-project/commit/416f6af9765cba2f9c45901b07e8d793f482f26d
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-08-14 (Mon, 14 Aug 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp

  Log Message:
  -----------
  AMDGPU: Remove special case folding of fma/mad

These just get replaced with an intrinsic now. This was also
introducing host dependence on the result since it relied on the
compiler choice to contract or not.


  Commit: 483cc21866255945c8230bb0de99857105b11ef9
      https://github.com/llvm/llvm-project/commit/483cc21866255945c8230bb0de99857105b11ef9
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-08-14 (Mon, 14 Aug 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp

  Log Message:
  -----------
  AMDGPU: Remove special case folding of sqrt


  Commit: d45022b094a0a00b52057b464902693bc4e2db76
      https://github.com/llvm/llvm-project/commit/d45022b094a0a00b52057b464902693bc4e2db76
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-08-14 (Mon, 14 Aug 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp
    M llvm/test/CodeGen/AMDGPU/simplify-libcalls.ll

  Log Message:
  -----------
  AMDGPU: Remove special case constant folding of divide

We should probably just swap this out for the fdiv, but that's what
the implementation is anyway.


  Commit: 1faa4797ca7cfef0b38e8404ae316e4728c2e67a
      https://github.com/llvm/llvm-project/commit/1faa4797ca7cfef0b38e8404ae316e4728c2e67a
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2023-08-14 (Mon, 14 Aug 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
    M llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
    M llvm/test/CodeGen/AMDGPU/llvm.exp.ll

  Log Message:
  -----------
  AMDGPU: Handle unsafe exp.f32 with denormal handling

I somehow missed this path when adding the new expansions. Saves a lot
of instructions for afn + IEEE.

https://reviews.llvm.org/D157867


Compare: https://github.com/llvm/llvm-project/compare/7d06f59b60e6...1faa4797ca7c