PR #27609 [MLAS][KleidiAI]Catlaw01/sgemm epilogue neon opt

[MLAS][KleidiAI]Catlaw01/sgemm epilogue neon opt #27609

hariharans29 merged 4 commits into microsoft:main from Laan33:catlaw01/sgemm-epilogue-neon-opt

Laan33 changed the title ~~Catlaw01/sgemm epilogue neon opt~~ [MLAS][KleidiAI] Catlaw01/sgemm epilogue neon opt 135 days ago

Laan33 changed the title ~~[MLAS][KleidiAI] Catlaw01/sgemm epilogue neon opt~~ [MLAS][KleidiAI]Catlaw01/sgemm epilogue neon opt 135 days ago

hariharans29 requested a review from

copilot-pull-request-reviewer 135 days ago

copilot-pull-request-reviewer commented on 2026-03-10

hariharans29 commented on 2026-03-10

Laan33 force pushed from becb89ff to 7b594fc0 134 days ago

hariharans29 commented on 2026-03-11

fix: Add ApplyBetaToC helper function and handle multiple batch sizes.

c2baed29

feat: Optimize ApplyAlphaBetaStrided for ARM NEON with vectorized path

086ef854

mlas: fix batched SGemm fast-path correctness and add short coverage

46f52d75

Remove conditional compilation for ARM_NEON in sgemm_kleidiai.cpp

f1d1717f

Laan33 force pushed from 2b591f7b to f1d1717f 127 days ago

hariharans29 approved these changes on 2026-03-18

hariharans29 enabled auto-merge (squash) 127 days ago

hariharans29 merged 3bb9e955 into main 127 days ago

Reviewers

hariharans29

Assignees

No one assigned

Labels

None yet

Milestone

No milestone