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++
[Engine] SYCL: make the prompt attention's query batch a switch, and measure it
Niko1221f7570e23e29b
diff --git a/sycl/src/prefill/prefill.cpp b/sycl/src/prefill/prefill.cppindex be06adfc5..c08464219 100644--- a/sycl/src/prefill/prefill.cpp+++ b/sycl/src/prefill/prefill.cpp@@ -734,7 +734,11 @@ struct Prefill::Impl {float* sel_scores = nullptr; // [sel_batch, max_blocks]int64_t sel_batch = 256, max_blocks = 0;float* attn_scratch = nullptr;- int64_t attn_batch = 32, cap = 0;+ int64_t attn_batch = [] { // SYCL port: how many queries the prompt path hands the FP32 attention kernel at once+ const char* v = std::getenv("STRATA_ATTN_BATCH"); // (the tensor-core kernel is opt-in and slower here)+ const long long n = v ? std::atoll(v) : 32; // 32 was picked for a vector-matrix kernel; more queries+ return (n >= 1 && n <= 1024) ? (int64_t) n : 32LL; // per launch give it more to overlap. Scratch grows too.+ }(), cap = 0;// MoEfloat *logits = nullptr, *w = nullptr, *GU = nullptr, *Dm = nullptr, *sgate = nullptr, *sup = nullptr,*shared = nullptr, *sg = nullptr;