llama.cpp
musa: define __CUDA_ARCH__ and fix the q8_0 dequantization kernel
#29508

Merged

musa: define __CUDA_ARCH__ and fix the q8_0 dequantization kernel #29508

CISC merged 1 commit into ggml-org:master from yeahdongcn:xd/musa-cuda-arch

github-actions added ggml

github-actions added CUDA

yeahdongcn force pushed from 13d5ab9e to 957a71e7 19 days ago

yeahdongcn force pushed from 957a71e7 to 00c9d563 19 days ago

yeahdongcn force pushed from 00c9d563 to 978cbb4d 19 days ago

yeahdongcn changed the title ~~musa: define __CUDA_ARCH__ for device passes~~ musa: define __CUDA_ARCH__ and fix the q8_0 and iq4_nl dequantization kernels 19 days ago

yeahdongcn requested a review from

CISC 19 days ago

yeahdongcn requested a review from

JohannesGaessler 19 days ago

yeahdongcn marked this pull request as ready for review 19 days ago

yeahdongcn requested a review 19 days ago

CISC commented on 2026-09-27

yeahdongcn closed this 18 days ago

yeahdongcn force pushed from 978cbb4d to 95887577 18 days ago

yeahdongcn reopened this 18 days ago

yeahdongcn marked this pull request as draft 18 days ago

yeahdongcn marked this pull request as ready for review 18 days ago

JohannesGaessler commented on 2026-09-29

yeahdongcn changed the title ~~musa: define __CUDA_ARCH__ and fix the q8_0 and iq4_nl dequantization kernels~~ musa: define __CUDA_ARCH__ and fix the q8_0 dequantization kernel 16 days ago

musa: define __CUDA_ARCH__ for device passes

bf76859b

yeahdongcn force pushed from cf51c4f6 to bf76859b 16 days ago

JohannesGaessler approved these changes on 2026-09-30

JohannesGaessler added merge ready

CISC approved these changes on 2026-09-30

CISC merged 272aad8b into master 15 days ago

Reviewers

CISC

JohannesGaessler

Assignees

No one assigned

Labels

ggml merge ready CUDA

Milestone

No milestone

llama.cpp musa: define __CUDA_ARCH__ and fix the q8_0 dequantization kernel #29508 Merged

musa: define __CUDA_ARCH__ and fix the q8_0 dequantization kernel #29508

llama.cpp
musa: define __CUDA_ARCH__ and fix the q8_0 dequantization kernel
#29508

Merged