diff options
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | 2021-03-25 10:01:52 +0000 |
|---|---|---|
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | 2021-03-25 10:18:44 +0000 |
| commit | 9fde88c3e21cc552af5fbb0b016fe186bad5c0f5 (patch) | |
| tree | 27190c337f4a0dedc0f7c1f386804a54067192c4 /clang/lib/Frontend/CompilerInvocation.cpp | |
| parent | d9abcdd9f471ab0d986197e94498a5cf90aa12ef (diff) | |
| download | llvm-9fde88c3e21cc552af5fbb0b016fe186bad5c0f5.zip llvm-9fde88c3e21cc552af5fbb0b016fe186bad5c0f5.tar.gz llvm-9fde88c3e21cc552af5fbb0b016fe186bad5c0f5.tar.bz2 | |
[X86][AVX] splitIntVSETCC - handle separate (canonicalized) SETCC operands
LowerVSETCC calls splitIntVSETCC after canonicalizing certain patterns, in particular (X & CPow2 != 0) -> (X & CPow2 == CPow2).
Unfortunately if we're splitting for AVX1/non-AVX512BW cases, we lose these canonicalizations as we call the split with the original SetCC node, and when the split nodes are later lowered in LowerVSETCC the patterns are lost behind extract_subvector etc. But if we pass the canonicalized operands for splitting we retain the optimizations.
Differential Revision: https://reviews.llvm.org/D99256
Diffstat (limited to 'clang/lib/Frontend/CompilerInvocation.cpp')
0 files changed, 0 insertions, 0 deletions