Vulkan - Xe Flash Attn / GEMM improvements from main PR#24408 - #2

Draft

XanderSC wants to merge 12 commits into

intel-gpu-feature

Commits on Jun 10, 2026

vulkan: add INTEL_PRE_XE2 arch enum and enable coopmat1 on Intel Xe-LPG Plus (1/3, Xe1-ARLH)

committed
7b24141
vulkan: add Intel Xe flash attention optimization kernels (2/3, Xe-LPG Plus/Xe2/Xe3)

committed
2ab4c94
vulkan: GEMM/Group GEMM optimizations and load-time weight compression for Intel MoE path (3/3, Xe-LPG Plus/Xe2/Xe3)

committed
f9e1fd0

Commits on Jun 12, 2026

vulkan: minor refine in FA decode kernels group max value of P matrix, refine Vulkan host code for temp buffer reuse and thread lock
sliu39
committed
77795e9

Commits on Aug 5, 2026