[PATCH] D36661: [AVX512] Don't switch unmasked subvector insert/extract instructions when AVX512DQI is enabled.

[Craig Topper via Phabricator via llvm-commits]

craig.topper created this revision.

There's no reason to switch instructions with and without DQI. It just creates extra isel patterns and test divergences.

There is however value in enabling the masked version of the instructions with DQI.

This required introducing some new multiclasses to enabling this splitting. If anyone has a better suggestion of how to do this I'm open to it.


https://reviews.llvm.org/D36661

Files:
  lib/Target/X86/X86InstrAVX512.td
  test/CodeGen/X86/avx512-cvt.ll
  test/CodeGen/X86/avx512-insert-extract.ll
  test/CodeGen/X86/avx512-trunc.ll
  test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll
  test/CodeGen/X86/compress_expand.ll
  test/CodeGen/X86/masked_gather_scatter.ll
  test/CodeGen/X86/stack-folding-fp-avx512.ll
  test/CodeGen/X86/stack-folding-int-avx512.ll
  test/CodeGen/X86/subvector-broadcast.ll
  test/CodeGen/X86/vector-compare-results.ll
  test/CodeGen/X86/vector-shuffle-512-v16.ll
  test/CodeGen/X86/vector-shuffle-avx512.ll
  test/CodeGen/X86/vector-shuffle-v1.ll
  test/CodeGen/X86/vector-trunc-math.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D36661.110912.patch
Type: text/x-patch
Size: 148290 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170814/f19b445b/attachment.bin>