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/GYRO.md: RTX 30-series setup line, RTX 4090 validation and the two-GPU notes
Laurent Zuijdwijkcce9b16a7219
diff --git a/docs/GYRO.md b/docs/GYRO.mdindex 36020d980..b033273d9 100644--- a/docs/GYRO.md+++ b/docs/GYRO.md@@ -1,7 +1,7 @@# Gyro (agentionai rotor quants) on Strata-Validated on Linux with an RTX 5090 (32 GB, CUDA 13.0, sm_120; 2026-10-02 and 2026-10-03) and an RTX 3090 (24 GB,-sm_86; 2026-10-03). A short end-to-end run on AMD (Strix Halo, gfx1151, ROCm) passed on 2026-10-03.+Validated on Linux with an RTX 5090 (32 GB, CUDA 13.0, sm_120; 2026-10-02 and 2026-10-03), an RTX 3090 (24 GB,+sm_86; 2026-10-03) and two RTX 4090s (24 GB each, sm_89; 2026-10-04, single and dual GPU). A short end-to-end run on AMD (Strix Halo, gfx1151, ROCm) passed on 2026-10-03.The targets are the published files of `agentionai/Qwen3.8-Flash-Next-Gyro-GGUF`, used unchanged:@@ -30,6 +30,7 @@ git clone -b rc1 https://github.com/agentionai/Strata && cd Strata```shtools/gyro_setup.sh --model S --backend cuda --cuda-arch 120 # RTX 50 seriestools/gyro_setup.sh --model S --backend cuda --cuda-arch 89 # RTX 40 series+tools/gyro_setup.sh --model S --backend cuda --cuda-arch 86 # RTX 30 series / RTX A6000tools/gyro_setup.sh --model S --backend hip --hip-arch gfx1201 # Radeon RX 9070 / AI PRO R9700 (experimental)```@@ -165,6 +166,16 @@ slower GPU, often on a slower link: read those rows as "it stays usable", not asThe decoded output was read and is fluent and correct; token-level parity against llama.cpp has not beenchecked. A real chat at 128k context worked through the web interface.+### Two GPUs (2026-10-04, 2× RTX 4090, no NVLink/P2P)++- **Layer split** (Strata splits the layers across both cards automatically) works with Gyro, including the+ idle-card prompt helper (`STRATA_PREFILL_HELP=1`). Repeated runs give identical output.+- **Decode with a second card** (`--peer-device 1` or `--expert-cache-device1 auto`) works and matches single-GPU output.+- **Prompt processing on the peer card** needs GPU-to-GPU access (NVLink or data-centre GPUs); without it Strata keeps+ the prompt on the first GPU and logs `no P2P`.++Single RTX 4090 smoke run (Gyro-S, MTP draft, greedy): code 107.5, prose 84.3, JSON 76.3 tokens/s.+## Known limits- **Not measured yet:** token parity against llama.cpp.