Skip to content

[SYCL] Flash Attention with XMX engine via oneDNN graph API (SDPA) on KV f16 for Xe2 ; Qwen3.6-27b-Q8_0 prefill speed up x1.21 at p=512 and x4.26 at p=80k - #25222

Merged
ggerganov merged 5 commits into
ggml-org:masterfrom
hmscider:sycl-fa-xmx-kv-f16
Jul 15, 2026