PR #14262 CUDA: mul_mat_v support for batch sizes > 1

CUDA: mul_mat_v support for batch sizes > 1 #14262

JohannesGaessler merged 3 commits into ggml-org:master from JohannesGaessler:cuda-mmv-bsgt1-9

CUDA: mul_mat_v support for batch sizes > 1

e22e345d

github-actions added Nvidia GPU

github-actions added ggml

use 64 bit math for initial offset calculation

2d24a9ce

fix mul_mat_id

60ce04c3

IMbackK approved these changes on 2025-06-22

JohannesGaessler merged defe2158 into master 93 days ago

Reviewers

IMbackK

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone