Setting up an Intel Arc GPU on Arch Linux for local LLM use (without ReBAR)
Getting an Intel Arc A750 to run local language models in an older NAS without Resizable BAR: the failed attempts, the Vulkan route, and what finally worked.
Stephen Bonner
Stephen Bonner · Personal blog
Machine learning, Linux, and things I’ve tried.
I’m Stephen, an ML researcher in Cambridge, UK. Here I write about hardware experiments, Linux, and what I learn along the way.
Pinned
Recent Posts
Using LVM with a cache for a root drive
Finding another use for an Intel Optane drive: setting up an LVM cache on Arch Linux and measuring what happens when you cache an already fast SSD.
Demystifying Intel’s GPU Software Stack for Local LLMs
Making sense of SYCL, oneAPI, Level Zero, and the layers between a local language model and an Intel GPU.