llama.cpp
CUDA: batch out_prod inner loop with cublasSgemmStridedBatched
#22651

Merged

CUDA: batch out_prod inner loop with cublasSgemmStridedBatched #22651

JohannesGaessler merged 3 commits into ggml-org:master from leonardHONG:cuda-out-prod-batched

CUDA: batch out_prod inner loop with cublasSgemmStridedBatched

cbde47de

CUDA: batch out_prod inner loop with cublasSgemmStridedBatched

5a5bbe4d

leonardHONG requested a review from

ggerganov 39 days ago

leonardHONG requested a review 39 days ago

github-actions added testing

github-actions added Nvidia GPU

github-actions added ggml

am17an approved these changes on 2026-05-04

JohannesGaessler approved these changes on 2026-05-04

CUDA: add cublasSgemmStridedBatched mapping for HIP and MUSA backends

a6284c91

leonardHONG requested a review from

IMbackK 37 days ago

JohannesGaessler approved these changes on 2026-05-07

JohannesGaessler merged 05ff59cb into master 35 days ago

Reviewers

JohannesGaessler

am17an

ggerganov

IMbackK

Assignees

No one assigned

Labels

testing Nvidia GPU ggml

Milestone

No milestone