diff options
| author | Craig Topper <craig.topper@intel.com> | 2019-06-30 06:46:37 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2019-06-30 06:46:37 +0000 |
| commit | 725a8a5dc43ff23640ca539b9b4c3f46bf549179 (patch) | |
| tree | d40b2e1b528120696b91e62e191b116e7eddd84c /llvm/lib/IR/Function.cpp | |
| parent | 4d0feb28ecbe177600d7081b2ad80c099732a970 (diff) | |
| download | llvm-725a8a5dc43ff23640ca539b9b4c3f46bf549179.zip llvm-725a8a5dc43ff23640ca539b9b4c3f46bf549179.tar.gz llvm-725a8a5dc43ff23640ca539b9b4c3f46bf549179.tar.bz2 | |
[X86] Custom lower AVX masked loads to masked load and vselect instead of selecting a maskmov+vblend during isel.
AVX masked loads only support 0 as the value for masked off elements.
So we need an extra blend to support other values. Previously we
expanded the masked load to two instructions with isel patterns.
With this patch we now insert the vselect during lowering and it
will be separately selected as a blend.
llvm-svn: 364718
Diffstat (limited to 'llvm/lib/IR/Function.cpp')
0 files changed, 0 insertions, 0 deletions