npu-inference

Open-source bring-up + verified matmul on the first-gen AMD XDNA1 (Phoenix/Hawk Point) NPU on Linux — the gen FastFlowLM/Lemonade skip. RyzenAI-npu1, mlir-aie/IRON, XRT.

Practical applications for the AMD XDNA NPU on Linux — webcam effects (working) + embeddings/audio/ggml roadmap. Built on open-xdna.

Tri-core NPU tunnel crack detection on Rockchip RK3588: packed-strips 3-in-1 inference (~2.9x), rknn_set_core_mask CORE_0/1/2, DMA-BUF full-chain zero-copy, consumes panorama DMA-BUF

RK3588 multi-thread NPU inference reusing official rknn_api + YOLOv8-pose in C, tri-core rknn_set_core_mask, DMA-BUF zero-copy

A collection of models, or source which get adapted or optimized to run on Huawei NPUs, this is a reference to help engineers use the super-cool NPUs better. 💅Led by engineers here in Middle East!

Architectural specification for a USB-C hardware appliance running decoupled IDE backends and local NPU AI coding assistants for air-gapped environments.

Agentic workflows ready, high-performance system and AI accelerator monitor for Rockchip SoCs (RK3576, RK3588) and RK1828/RK182x coprocessors. Written in Rust.