llama.cpp
[Metal/Apple Silicon Performance] Disable K-quant mul_mv_ext for M3 Pro in lieu of regular mul_mv
#27776
Open

[Metal/Apple Silicon Performance] Disable K-quant mul_mv_ext for M3 Pro in lieu of regular mul_mv #27776

DavidLandup0
DavidLandup0 metal: turn K-quant mul_mv_ext off on M3 Pro
eaecb55a
DavidLandup0 metal: rename to use_kq_mv_ext
dbf28aa0
DavidLandup0 metal: extract the boolean logic into ggml-metal-tuning.cpp
1e1a7763
DavidLandup0 DavidLandup0 requested a review 1 day ago
github-actions github-actions added ggml
github-actions github-actions added Apple Metal
DavidLandup0 DavidLandup0 changed the title [Metal/Apple Silicon Performance] Disable mul_mv_ext for M3 Chips [Metal/Apple Silicon Performance] Disable K-quant mul_mv_ext for M3 Pro in lieu of regular mul_mv 1 day ago

Login to write a write a comment.

Login via GitHub

Reviewers
No reviews
Assignees
No one assigned
Labels
Milestone