npu-monitor: a system tray that knows what hardware each box actually has

Fri, 21 Aug 2026 00:00:00 +0000

Every multi-host monitoring tool I tried assumed homogeneous hardware: one GPU, maybe two, same on every machine. My LAN doesn’t work like that — one box has an AMD APU iGPU plus a Ryzen AI NPU, another has an AMD R9700 and an NVIDIA RTX 5060 Ti. So npu-monitor treats the device list as dynamic: each host reports its own real silicon and the tray renders whatever arrives.

The numbers come from the actual kernel interfaces, not screen-scraping: ROCm sysfs (gpu_busy_percent) for AMD cards, nvidia-smi for CUDA, /proc/stat deltas for CPU, and a raw DRM_IOCTL_AMDXDNA_GET_INFO ioctl against /dev/accel/accel0 — using the real 168-byte sensor struct from the kernel UAPI — for the NPU’s utilization columns.

wimpy-setup: a dual-GPU inference box you can rebuild from scripts

Sun, 28 Jun 2026 00:00:00 +0000

This is not a guide. It’s the actual, ordered, idempotent script set that builds my inference server — shared because rebuilding from memory is how homelabs die. Wimpy is a Ryzen 7 7700 with two GPUs serving LLM inference at the same time: an AMD Radeon AI PRO R9700 (32GB, ROCm) and an NVIDIA RTX 5060 Ti (16GB, CUDA), behind one llama-swap endpoint.

The dual-GPU part is the reason this repo exists. HIP and CUDA backends of llama.cpp share library filenames, so they can’t coexist in one prefix — there are two isolated builds (/usr/local for ROCm, /opt/llama-cuda for CUDA) and one llama-swap config with two non-exclusive groups. A -cuda model entry is the same GGUF as its AMD twin with a different binary and device pin, so either card can serve it. GPUs are pinned by UUID (the Ryzen iGPU also enumerates as a ROCm device and will happily steal your workloads if you let it).

Linux on AiSlackin

npu-monitor: a system tray that knows what hardware each box actually has

wimpy-setup: a dual-GPU inference box you can rebuild from scripts