Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
C++
docs/sycl: PF_XMX=2 Alchemist default approved by the owner 2026-10-18 (comment and INTEL.md heading only)
Niko1221cfe65f8d1ad3
diff --git a/docs/INTEL.md b/docs/INTEL.mdindex 1cbe0c902..dc1e6542b 100644--- a/docs/INTEL.md+++ b/docs/INTEL.md@@ -759,7 +759,7 @@ Measured on an Arc A750 (8 GB, `i915`, PCIe 4.0) with the Flash-Next IQ3_XXS in8192 --prefill-repeat 5` for the time of both kernels. Not taken: the same kernel with its state in SLM (no registerspills, the same bits, ~4x slower: 13.7 vs 3.1-3.3 ms at T=1000).-### Prompt speed on Alchemist: the grouped XMX expert launch is the default (0.1.48, `STRATA_PF_XMX`)+### Prompt speed on Alchemist: the grouped XMX expert launch is the default (0.1.48, `STRATA_PF_XMX`; owner approved 2026-10-18)The A750's prompt path decodes and multiplies a group of experts in one fused launch on the XMX units (`STRATA_PF_XMX=2`, `sycl/src/prefill/xmx_moe.dp.cpp`)instead of dequantizing each expert to FP16 and calling oneMKL. It is the default on Alchemist cards; `STRATA_PF_XMX=0` turns it off and `=1` is thediff --git a/sycl/src/prefill/xmx_moe.dp.cpp b/sycl/src/prefill/xmx_moe.dp.cppindex 8cbdecf72..c645e1bf3 100644--- a/sycl/src/prefill/xmx_moe.dp.cpp+++ b/sycl/src/prefill/xmx_moe.dp.cpp@@ -150,7 +150,7 @@ int mode() {// sub-group 16 (Xe2, Battlemage).static const int v = [] {const char* e = std::getenv("STRATA_PF_XMX");- // 0.1.48: mode 2 (the fused grouped XMX launch) is the default on Arc Alchemist (A-series): A750 prompts +30..39% (4K), +32..35% (1-2K tokens),+ // 0.1.48 (owner approved 2026-10-18): mode 2 (the fused grouped XMX launch) is the default on Arc Alchemist (A-series): A750 prompts +30..39% (4K), +32..35% (1-2K tokens),// first-token KL 0.002 against a 0.0005 floor, HumanEval / reasoning / agent / CJK / needles inside the run-to-run spread; Battlemage is// 7-11% slower with it, so it stays off there. STRATA_PF_XMX=0 turns it off, 1 is the dequantize + grouped XMX variant.const bool alchemist = strata::intel_gpu_gen(strata::q_of(nullptr)->get_device()) == strata::IntelGpuGen::Alchemist;