The AMDXDNA open-source Linux kernel accelerator driver is seeing a fair amount of work for better supporting current-generation NPU3 neural processing units. The AMD NPU3 is what's found in current AMD Ryzen AI 300 / Ryzen AI 300 PRO series SoCs.

AMD engineer David Zhang has been working on the AMDXDNA driver to handle kernel-mode command submission for NPU3 hardware as well as improving system and run-time power management support like for suspend/resume. He explained in the latest revision of the patches:

"This patch series extends the amdxdna accelerator driver to support AMD AIE4 (NPU3) platforms with kernel-mode command submission (KMQ), device telemetry queries, comprehensive system and runtime power management, and NPU3 classic device support.

Previously, AIE4 devices relied on userspace doorbell mapping for command dispatch. This series establishes kernel-mode submission as the default execution path, providing robust fence management, BO reservation locking, and queue lifecycle handling. Additionally, it implements full system suspend/resume and runtime power management across Physical Functions (PF), Virtual Functions (VF), and Classic devices, along with SR-IOV PM dependency linking and userspace runtime compatibility stubs."

The current set of 21 patches lay out the kernel-mode command submission, enabling new clock / frequency / information queries, suspend and resume support, and other improvements for AIE4/NPU3 hardware.

More details for those interested in this pending AMDXDNA Linux driver work can find the latest patches on the LKML.