[PATCH] D35978: [AVX-512] Don't use unmasked VMOVDQU8/16 for 8-bit or 16-bit element stores even when BWI instructions are supported. Always use VMOVDQA32/VMOVDQU32.

[Craig Topper via Phabricator via llvm-commits]

craig.topper updated this revision to Diff 109014.
craig.topper added a comment.

Fixed the execution domain issue in r309632.

Changed AVX3 to AVX512 in r309625.

Added common prefix to avx512-insert-extract.ll in r309629


https://reviews.llvm.org/D35978

Files:
  lib/Target/X86/X86InstrAVX512.td
  test/CodeGen/X86/avg.ll
  test/CodeGen/X86/avx512-insert-extract.ll
  test/CodeGen/X86/avx512-insert-extract_i1.ll
  test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll
  test/CodeGen/X86/avx512bw-mov.ll
  test/CodeGen/X86/avx512bwvl-mov.ll
  test/CodeGen/X86/subvector-broadcast.ll
  test/CodeGen/X86/x86-interleaved-access.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D35978.109014.patch
Type: text/x-patch
Size: 25790 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170731/64d16c7c/attachment.bin>