PyTorch with Vulcan backend

nirs · May 30, 2025, 10:52pm

I found docs for building Vulcan backend:

I want to use this to play with PyTorch in minikube krunkit driver on macOS:

github.com/kubernetes/minikube

WIP: krunkit driver

master ← nirs:krunkit

opened 07:11PM - 23 May 25 UTC

nirs

+1074 -886

krunkit is a tool to launch configurable virtual machines using the libkrun plat…form, optimized for GPU accelerated virtual machines and AI workloads on Apple silicon. It is mostly compatible with vfkit; the driver is a simplified copy of the vfkit driver. To get accelerated gpu the arch iso was updated to longterm kernel 6.6.92 and Virtio GPU was enabled. Limitations: - Only one machine can be created since we use the same port for krunkit --restful-uri. This should be fix to use an unused port or use unix socket when supported[1] [1] https://github.com/containers/krunkit/issues/47 Status: - ✅ The machine boots - ✅ Networking with vment-helper works - ✅ /dev/dri/ is available - ❌ qemu does does not boot with kernel 6.6 Based on these PRs for testing: - #20832 - #20831 Fixes #20803

It seems that I need to build it in a container, and copy the build results to a demo container. I wonder if the built artifacts are contained and easy to copy.

Or maybe we already have an example how to create a container with pythorch that I can reuse?