[PATCH] D111864: [AMDGPU] Avoid redundant calls to numBits in AMDGPUCodeGenPrepare::replaceMulWithMul24().
Jay Foad via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Fri Oct 15 02:04:42 PDT 2021
foad added inline comments.
================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp:512
+ } else if (ST->hasMulI24() &&
+ Size >= 24 && // Types less than 24-bit should be treated as u24.
+ (LHSBits = numBitsSigned(LHS, Size)) < 24 &&
----------------
I realize this check was already in isI24, but I don't see the point of it. I think the only effect it has is: on a machine that has mul_i24 but not mul_u24, we will fail to do this optimization, for no good reason.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D111864/new/
https://reviews.llvm.org/D111864
More information about the llvm-commits
mailing list