Page MenuHomePhabricator

[X86][AVX512]: Adding full coverage of MC encoding for the AVX512 isa sets (w/o AVX512F).<NFC>
Needs ReviewPublic

Authored by tianqing on Dec 3 2017, 9:22 AM.

Details

Summary

NFC.
Adding MC regressions tests to cover the AVX512 (without AVX512F) ISA sets.
This patch is part of a larger task to cover MC encoding of all X86 ISA Sets.
See revision: https://reviews.llvm.org/D39952

Diff Detail

Repository
rL LLVM

Event Timeline

gadi.haber created this revision.Dec 3 2017, 9:22 AM

I don't see any tests for {z} masking. And most instructions only test with a {%kX} mask. There's no testing of unmasked instructions eitehr.

Craig, following your comment I added the zero masked and non masked instructions. However, the diff file is too large to get uploaded by the phabricator.
Consequently, I will have to split the review into multiple reviews.

Added zero masked and non-masked instructions following Craig's comment

gadi.haber retitled this revision from [X86][AVX512]: Adding full coverage of MC encoding for the AVX512 isa set.<NFC> to [X86][AVX512]: Adding full coverage of MC encoding for the AVX512 isa sets (w/o AVX512F).<NFC>.Dec 13 2017, 4:41 AM
gadi.haber edited the summary of this revision. (Show Details)
craig.topper added inline comments.Jan 11 2018, 5:52 PM
test/MC/X86/AVX512BW_128N-64.s
4

This is a legacy SSE instruction. It shoudldn't be here.

12

This is testing the VEX encoded version not the EVEX encoded version. The encoding for EVEX should start with 0x62. I believe this problem exists in a lot of the tests without masking.

tianqing commandeered this revision.Jan 9 2019, 6:06 PM
tianqing added a reviewer: gadi.haber.

@tianqing Are you still looking at this?

@RKSimon Yes, there is some non-EVEX encoding inside, so I'm trying to regenerate it.

@tianqing It might make sense to split this patch down into per-ISA patches?

Herald added a project: Restricted Project. · View Herald TranscriptSep 1 2019, 4:50 AM