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++
DeepSeek HIP: the matrix-core prompt attention (DS4_FAST_ATTN) is on by default on gfx1200/gfx1201 only; DS4_FAST_ATTN=0 is the off switch, gfx11 and gfx1151 stay scalar unless =1
Niko12211f079f9d13a9
diff --git a/src/dsv41/ds4_kernels.cu b/src/dsv41/ds4_kernels.cuindex ff2aefeb2..96d2474aa 100644--- a/src/dsv41/ds4_kernels.cu+++ b/src/dsv41/ds4_kernels.cu@@ -14,6 +14,7 @@#include <atomic>#include <cfloat>#include <cstdlib>+#include <cstring>#include <map>#include <mutex>#include <cstdio>@@ -2025,15 +2026,40 @@ bool split_enabled() {}();return on;}-// DS4_FAST_ATTN=1: the chunked prompt's packed-KV attention on the matrix cores (sparse_attn_wmma_kernel). Opt-in: the+// DS4_FAST_ATTN: the chunked prompt's packed-KV attention on the matrix cores (sparse_attn_wmma_kernel). The// QK/PV sums are the matrix cores' order, so the outputs differ from the scalar kernel's in the last places (the owner// approved the trade for the prefill; the decode path stays on the scalar kernels).+// 0.2.7: ON by default on gfx1200 / gfx1201 (RDNA4: the R9700, RX 9000), where it was measured and gated (KL 0.0003 against the+// 0.006 gate, ids identical); gfx11 and gfx1151 (Strix Halo) are untested and keep the scalar kernels. DS4_FAST_ATTN=0 is the+// off switch (exactly the 0.2.6 arithmetic), DS4_FAST_ATTN=1 turns it on elsewhere. Decided per device (the first call there).+#if defined(STRATA_USE_HIP)+bool device_is_gfx12() {+ static std::atomic<int> cache[64]; // 0 unknown, 1 no, 2 yes+ int dev = 0;+ cudaGetDevice(&dev);+ if (dev < 0 || dev >= 64) return false;+ int c = cache[dev].load();+ if (c == 0) {+ cudaDeviceProp prop{};+ const bool yes = cudaGetDeviceProperties(&prop, dev) == cudaSuccess &&+ (std::strncmp(prop.gcnArchName, "gfx1200", 7) == 0 || std::strncmp(prop.gcnArchName, "gfx1201", 7) == 0);+ c = yes ? 2 : 1;+ cache[dev].store(c);+ }+ return c == 2;+}+#endifbool fast_attn_enabled() {- static const bool on = [] {+ static const int env = [] {const char* e = std::getenv("DS4_FAST_ATTN");- return e != nullptr && std::atoi(e) != 0;+ return e == nullptr ? -1 : (std::atoi(e) != 0 ? 1 : 0);}();- return on;+ if (env >= 0) return env == 1;+#if defined(STRATA_USE_HIP)+ return device_is_gfx12();+#else+ return false;+#endif}attnsplit::AttnSplitScratch split_scratch(cudaStream_t s, int heads) {static std::mutex mu;