llama.cpp
[SYCL] add DPAS flash attention kernel for quantized KV cache
#31125

Merged

[SYCL] add DPAS flash attention kernel for quantized KV cache #31125

arthw merged 4 commits into ggml-org:master from Torchit1:sycl-fa-q4-xmx-up

sycl : add DPAS flash attention kernel for quantized KV cache

105df3aa

Torchit1 requested a review from

ggerganov 14 days ago

Torchit1 requested a review 14 days ago

github-actions added documentation

github-actions added testing

github-actions added ggml

github-actions added SYCL

Titaniumtown requested changes on 2026-10-25

sycl : 192 head dropped, Wildcat Lake added to docs

790c1254

Titaniumtown commented on 2026-10-25

sycl : stricter checks from review

befe7a60

Titaniumtown approved these changes on 2026-10-25

Torchit1 changed the title ~~sycl : add DPAS flash attention kernel for quantized KV cache~~ [SYCL] add DPAS flash attention kernel for quantized KV cache 13 days ago

Titaniumtown requested a review from

arthw 12 days ago

arthw commented on 2026-10-27

[SYCL] : rename XMX macro, add GGML_SYCL_SUPPORT_XMX option, document…

e81b227c

arthw approved these changes on 2026-10-28

arthw merged f7c01c52 into master 10 days ago

ggerganov commented on 2026-10-29

Reviewers

arthw

Titaniumtown

ggerganov

Assignees

No one assigned

Labels

documentation testing ggml SYCL

Milestone

No milestone

llama.cpp [SYCL] add DPAS flash attention kernel for quantized KV cache #31125 Merged

[SYCL] add DPAS flash attention kernel for quantized KV cache #31125

llama.cpp
[SYCL] add DPAS flash attention kernel for quantized KV cache
#31125

Merged