[all-commits] [llvm/llvm-project] 8d2e61: [AMDGPU] Calculate getDivNumBits' AtLeast using bi...

[choikwa via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 8d2e611802d5f3bdd681d308ceb293e5ace8a894
      https://github.com/llvm/llvm-project/commit/8d2e611802d5f3bdd681d308ceb293e5ace8a894
  Author: choikwa <5455710+choikwa at users.noreply.github.com>
  Date:   2025-01-07 (Tue, 07 Jan 2025)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp
    M llvm/test/CodeGen/AMDGPU/sdiv64.ll
    M llvm/test/CodeGen/AMDGPU/srem64.ll

  Log Message:
  -----------
  [AMDGPU] Calculate getDivNumBits' AtLeast using bitwidth (#121758)

Previously in shrinkDivRem64, it used fixed value 32 for AtLeast which
meant that <64bit divisions would be rejected from shrinking since logic
depended only on number of sign bits. I.e. 'idiv i48 %0, %1' would
return 24 for number of sign bits if %0,%1 both had 24 division bits,
and was rejected.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications