PR #8075 CUDA: use MMQ instead of cuBLAS by default

CUDA: use MMQ instead of cuBLAS by default #8075

JohannesGaessler merged 1 commit into ggml-org:master from JohannesGaessler:cuda-mmq-default

github-actions added build

github-actions added Nvidia GPU

github-actions added ggml

slaren commented on 2024-06-24

JohannesGaessler force pushed from cc029763 to 5479853c 1 year ago

mofosyne added Review Complexity : Medium

slaren approved these changes on 2024-06-24

CUDA: use MMQ instead of cuBLAS by default

61f3cb6e

JohannesGaessler force pushed from 5479853c to 61f3cb6e 1 year ago

JohannesGaessler merged a818f302 into master 1 year ago

Reviewers

slaren

Assignees

No one assigned

Labels

build Nvidia GPU Review Complexity : Medium ggml

Milestone

No milestone