This is an archive of the discontinued LLVM Phabricator instance.

AVX512: Fix truncate v32i8 to v32i1 lowering implementation.
ClosedPublic

Authored by igorb on Jan 25 2016, 6:48 AM.

Details

Summary

AVX512: Fix truncate v32i8 to v32i1 lowering implementation.
bug bug 26264 .Enable truncate 128/256bit packed byte/word with AVX512BW but without AVX512VL, use 512bit instructions.

Diff Detail

Repository
rL LLVM

Event Timeline

igorb updated this revision to Diff 45861.Jan 25 2016, 6:48 AM
igorb retitled this revision from to AVX512: Fix truncate v32i8 to v32i1 lowering implementation..
igorb updated this object.
igorb added a reviewer: delena.
igorb set the repository for this revision to rL LLVM.
igorb added a subscriber: llvm-commits.
igorb updated this revision to Diff 46231.Jan 28 2016, 12:04 AM

implement truncate truncate 128/256bit using 512bit instructions id TD file.

delena added inline comments.Jan 28 2016, 12:47 AM
test/CodeGen/X86/masked_memop.ll
4

Please put this check it in a separate test.

igorb updated this revision to Diff 46246.Jan 28 2016, 1:32 AM
igorb marked an inline comment as done.
delena accepted this revision.Jan 28 2016, 3:43 AM
delena edited edge metadata.

You can fix the small code style issue and commit.

lib/Target/X86/X86ISelLowering.cpp
13592

remove else after return

This revision is now accepted and ready to land.Jan 28 2016, 3:43 AM
This revision was automatically updated to reflect the committed changes.
igorb marked an inline comment as done.Jan 28 2016, 5:24 AM

Many Thanks for your valuable comments and suggestions.