diff options
| author | Craig Topper <craig.topper@gmail.com> | 2020-02-27 20:53:17 -0800 |
|---|---|---|
| committer | Craig Topper <craig.topper@gmail.com> | 2020-03-01 16:38:08 -0800 |
| commit | b6e2796114d08aadfabe8c889b5d96e6bc4f5e0e (patch) | |
| tree | 5d1ddce3f0a714e85f7a0d23b7e98c718ef0d937 /llvm/lib/Bitcode/Writer/BitcodeWriter.cpp | |
| parent | 211fb91f1067ecdf7c0b8a019bcf76554d813129 (diff) | |
| download | llvm-b6e2796114d08aadfabe8c889b5d96e6bc4f5e0e.zip llvm-b6e2796114d08aadfabe8c889b5d96e6bc4f5e0e.tar.gz llvm-b6e2796114d08aadfabe8c889b5d96e6bc4f5e0e.tar.bz2 | |
[X86][TwoAddressInstructionPass] Teach tryInstructionCommute to continue checking for commutable FMA operands in more cases.
Previously we would only check for another commutable operand if the first commute was an aggressive commute.
But if we have two kill operands and neither is tied to the def at the start, we should consider both operands as the one to use as the new def.
This improves the loop in the fma-commute-loop.ll test. This test is derived from a post from discourse here https://llvm.discourse.group/t/unnecessary-vmovapd-instructions-generated-can-you-hint-in-favor-of-vfmadd231pd/582
Differential Revision: https://reviews.llvm.org/D75016
Diffstat (limited to 'llvm/lib/Bitcode/Writer/BitcodeWriter.cpp')
0 files changed, 0 insertions, 0 deletions