llama.cpp
Vulkan: Support fp32 accumulator in quantized matmul to fix GLM4-32B incoherence
#13607

Merged

Vulkan: Support fp32 accumulator in quantized matmul to fix GLM4-32B incoherence #13607

0cc4m merged 1 commit into master from 0cc4m/vulkan-mmq-fp32-acc-glm4

Vulkan: Add f32 accumulator support to quantized mul mat to fix GLM4 …

35d675d8

0cc4m requested a review from

jeffbolznv 1 year ago

github-actions added Vulkan

github-actions added ggml

jeffbolznv approved these changes on 2025-05-18

0cc4m marked this pull request as ready for review 1 year ago

0cc4m merged 8960efd0 into master 1 year ago

0cc4m deleted the 0cc4m/vulkan-mmq-fp32-acc-glm4 branch 1 year ago

Reviewers

jeffbolznv

Assignees

No one assigned

Labels

Vulkan ggml

Milestone

No milestone