| commit | b078fabb56e34115e55357c319d589b9455dd189 | |
| author | Julian Seward <jseward@acm.org> | |
| Sat, 22 Dec 2018 06:23:00 +0000 (22 07:23 +0100) | ||
| committer | Julian Seward <jseward@acm.org> | |
| Sat, 22 Dec 2018 06:23:00 +0000 (22 07:23 +0100) | ||
| tree | b965367521fee5bcfafe47961117f3327b651c36 | treesnapshot (tar.gz zip) |
| parent | 6cb6bdbd0a38e9b5f5c4f676afb72a23b6bfb1b5 | commitdiff |
amd64 pipeline: generate much better code for pshufb mm/xmm/ymm. n-i-bz.
pshufb mm/xmm/ymm rearranges byte lanes in vector registers. It's fairly
widely used, but we generated terrible code for it. With this patch, we just
generate, at the back end, pshufb plus a bit of masking, which is a great
improvement.
pshufb mm/xmm/ymm rearranges byte lanes in vector registers. It's fairly
widely used, but we generated terrible code for it. With this patch, we just
generate, at the back end, pshufb plus a bit of masking, which is a great
improvement.
12 files changed: