diff options
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | 2018-08-29 10:51:08 +0000 |
|---|---|---|
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | 2018-08-29 10:51:08 +0000 |
| commit | 6b9bf7ecbc152d5941712f967cd93d1c0e2042a7 (patch) | |
| tree | f186aae5f625f35467587cd64022ec269a8b3ad1 /llvm/lib/Transforms/Utils | |
| parent | edc318166fc9b2882fdab0d004bc3bdb598197d6 (diff) | |
| download | bcm5719-llvm-6b9bf7ecbc152d5941712f967cd93d1c0e2042a7.tar.gz bcm5719-llvm-6b9bf7ecbc152d5941712f967cd93d1c0e2042a7.zip | |
[X86][AVX] Prefer VPBLENDW+VPBLENDD to VPBLENDVB for v16i16 blend shuffles
Noticed while looking at D49562 codegen - we can avoid a large constant mask load and a slow VPBLENDVB select op by using VPBLENDW+VPBLENDD instead.
TODO: As discussed on the patch, we should investigate adding VPBLENDVB handling to target shuffle combining as well, that will allow us to extend this to VPBLENDW+VPBLENDW+VPBLENDD.
Differential Revision: https://reviews.llvm.org/D50074
llvm-svn: 340913
Diffstat (limited to 'llvm/lib/Transforms/Utils')
0 files changed, 0 insertions, 0 deletions
