[hedgehog]

This is SSH into a remote Linux VM that has the GPU attached.

[jackb4040]

This. I was ready to be impressed if author had written some software abstraction layer that wrapped a remote GPU in a virtual device that could be used by existing programs running on the local CPU. But there is no novel software at all, they're just remoting into WSL on the other machine.

[syntaxing]

Same, I thought it was about some cool network tunneling that made the GPU look like native hardware on the Linux host.

[0x20cowboy]

Check out exo if you haven’t already https://github.com/exo-explore/exo

[jackb4040]

Thank you! Looks cool

| exo provides multiple API-compatible interfaces for maximum compatibility with existing tools: | | OpenAI Chat Completions API - Compatible with OpenAI clients | Claude Messages API - Compatible with Anthropic's Claude format | OpenAI Responses API - Compatible with OpenAI's Responses format | Ollama API - Compatible with Ollama and tools like OpenWebUI

So a bit higher-level than a virtual device, but yes still pretty neat.