projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 5bf32a0) | patch
[AMDGPU] Avoid redundant calls to numBits in AMDGPUCodeGenPrepare::replaceMulWithMul24().
authorAbinav Puthan Purayil <abinav.puthanpurayil@amd.com>
Fri, 15 Oct 2021 03:59:58 +0000 (09:29 +0530)
committerAbinav Puthan Purayil <abinav.puthanpurayil@amd.com>
Fri, 15 Oct 2021 14:19:44 +0000 (19:49 +0530)
commitde3038400b163d91dca743210ce9b10001411208
tree45a907c3f308388297f68c3214d6e116a8ae8ae5tree | snapshot
parent5bf32a0e4841d56b481344fb4d074ac96a854261commit | diff
[AMDGPU] Avoid redundant calls to numBits in AMDGPUCodeGenPrepare::replaceMulWithMul24().

The isU24() and isI24() calls numBits to make its decision. This change
replaces them with the internal numBits call so that we can use its
result for the > 32 bit width cases.

Differential Revision: https://reviews.llvm.org/D111864
llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp diff | blob | history
Domain: System / Toolchain;