llama.cpp
vulkan: handle quantize_q8_1 overflowing the max workgroup count
#18515

Merged

vulkan: handle quantize_q8_1 overflowing the max workgroup count #18515

0cc4m merged 3 commits into ggml-org:master from jeffbolznv:issue_18510

vulkan: handle quantize_q8_1 overflowing the max workgroup count

9b1b9618

jeffbolznv requested a review from

ggerganov 85 days ago

jeffbolznv requested a review from

0cc4m 85 days ago

jeffbolznv commented on 2025-12-31

github-actions added testing

github-actions added Vulkan

github-actions added ggml

vulkan: Fix small tile size matmul on lavapipe

cdf2427a

jeffbolznv commented on 2026-01-01

fix mul_mat_id failures

cd53729e

0cc4m approved these changes on 2026-01-02

0cc4m merged b37124d2 into master 80 days ago

Reviewers

0cc4m

ggerganov

Assignees

No one assigned

Labels

testing Vulkan ggml

Milestone

No milestone