| commit | 40847a4c84ddbe2666ae1751a0aa2fdf4549c65d | [log] [tgz] |
|---|---|---|
| author | Sanjay Patel <spatel@rotateright.com> | Wed Mar 09 22:12:08 2016 +0000 |
| committer | Sanjay Patel <spatel@rotateright.com> | Wed Mar 09 22:12:08 2016 +0000 |
| tree | dcd0b41efc3cc2c80c60615614d346d798ce5c0a | |
| parent | 352b0048bacc403cad2df26cbf5e68ca92269c6d [diff] |
[x86, AVX] optimize masked loads with constant masks Instead of a variable-blend instruction, form a blend with immediate because those are always cheaper. Differential Revision: http://reviews.llvm.org/D17899 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@263067 91177308-0d34-0410-b5e6-96231b3b80d8