llama.cpp
ggml-cpu: add x86 VNNI Q2_0 dot product -- 3x speed improvement for VNNI-compatible CPUs
#26348
Merged

ggml-cpu: add x86 VNNI Q2_0 dot product -- 3x speed improvement for VNNI-compatible CPUs #26348

Michionlion
Michionlion Michionlion requested a review from ggerganov ggerganov 108 days ago
github-actions github-actions added ggml
ggml-gh-bot
Michionlion
Michionlion Michionlion force pushed to e88e6319 108 days ago
am17an
Michionlion
Michionlion
Michionlion Michionlion force pushed from 4cc28ba1 to 915ffcfb 88 days ago
AlexGabbia
maicomferre
CISC
lxe
CISC
Michionlion ggml-cpu: use VNNI for Q2_0 dot products
cdecb2fe
Michionlion Michionlion force pushed to cdecb2fe 26 days ago
Michionlion ggml-cpu: use VBMI for Q2_0 unpacking
9ff97f5a
Michionlion
CISC
CISC commented on 2026-10-21
CISC CISC requested a review 26 days ago
Michionlion Update ggml/src/ggml-cpu/arch/x86/quants.c
58133076
SvenMeyer
volkjohn
Michionlion
am17an
am17an approved these changes on 2026-11-15
CISC
CISC approved these changes on 2026-11-15
CISC CISC merged b54752df into master 21 hours ago
Michionlion Michionlion deleted the fix/q2-0-x86-vnni branch 15 hours ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone