perf(avx512vl): native EVEX int64 shift, signed rotr, compress/expand#1358
Draft
DiamonDinoia wants to merge 1 commit into
Draft
perf(avx512vl): native EVEX int64 shift, signed rotr, compress/expand#1358DiamonDinoia wants to merge 1 commit into
DiamonDinoia wants to merge 1 commit into
Conversation
Add pure-VL overrides so these stop falling back to AVX2/SSE:
- int64 signed >> -> vpsraq/vpsravq (unsigned/32-bit unchanged)
- signed rotr -> vprorvq/vprolq (drop is_unsigned guard, mirror rotl)
- compress/expand -> EVEX forms; 8/16-bit fall through to common{}
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
avx512vl_{128,256}inherit AVX2/SSE codegen for un-overridden kernels. Adds three pure-VL overrides (no DQ/CD, no VBMI2), each one EVEX op:>>(signed):vpsraq/vpsravq(was ~5–10-instr emulation); unsigned/32-bit keep the inherited path.rotr:vprorvq/vprolq(dropped theis_unsignedguard to mirrorrotl).VPCOMPRESSD/Q,VCOMPRESSPS/PDand theVPEXPAND*forms for 32/64-bit int + float/double.8/16-bit
compress/expand(VPCOMPRESSB/VPCOMPRESSW) require AVX512_VBMI2 and are intentionally deferred — they fall through to thecommon{}fallback here and will be added later alongside newavx512vbmi2_{128,256}arch types.Depends on #1357 (its swizzle fix unblocks the
avx512vl_256test build).