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 engine (AMD): the staging ring from each device's room (not the summed room)
Niko12213b2399456840
diff --git a/src/dsv41/ds4_model.cu b/src/dsv41/ds4_model.cuindex e81458ede..30c06b39b 100644--- a/src/dsv41/ds4_model.cu+++ b/src/dsv41/ds4_model.cu@@ -2913,8 +2913,9 @@ struct Model::Impl {// at 64-400 MB, unchanged at 16). When they do not fit, the stream IS the prompt's expert stage, and its depth// is what decides how long the tiles wait: on the 4x R9700 8K prompt a 16 MB ring left 1.7 s of DMA waits in a// 5.2 s expert stage, and 1.2-3 GB of ring measured 632 -> 678-684 ppts (+7%) for about -2% decode (off a- // fifth more CPU-lane experts). So: a share of the room, bounded, rather than the fixed 400 MB of 0.2.6.- const size_t deep = (size_t) std::min(2000.0, std::max(400.0, room / 1e9 * 0.05 * 1000.0)); // (MB)+ // fifth more CPU-lane experts). So: a share of each device's room, bounded, rather than the fixed 400 MB.+ const double per_dev = devs.empty() ? 0.0 : room / (double) devs.size();+ const size_t deep = (size_t) std::min(2000.0, std::max(400.0, per_dev / 1e9 * 0.08 * 1000.0)); // (MB)const size_t mb = share >= 0.98 ? 16 : deep;if (o.verbose)std::fprintf(stderr, "ds4: staging ring: experts %.1f GB, VRAM room %.1f GB, dense %.1f GB -> %d MB\n", exp_b / 1e9,