llama.cpp
metal : skip the empty half of the mul_mm_id token tile, load iq2/iq3 codebooks as uint32
#28301

Merged

metal : skip the empty half of the mul_mm_id token tile, load iq2/iq3 codebooks as uint32 #28301

ggerganov merged 1 commit into ggml-org:master from masterFoad:metal-mmid-halftile-pr

masterFoad requested a review from

ggerganov 66 days ago

masterFoad requested a review 66 days ago

masterFoad marked this pull request as draft 66 days ago

github-actions added testing

github-actions added ggml

github-actions added Apple Metal

ggerganov assigned

ggerganov 63 days ago

masterFoad marked this pull request as ready for review 60 days ago

masterFoad force pushed 58 days ago

ggerganov commented on 2026-09-11

ggerganov approved these changes on 2026-09-11

metal : skip the empty half of the mul_mm_id token tile

e80a3226

masterFoad force pushed to e80a3226 58 days ago

ggerganov merged 5bda51bf into master 58 days ago

Reviewers

ggerganov

Assignees

ggerganov

Labels

testing ggml Apple Metal

Milestone

No milestone