[rektide]

Beautiful effort. I'm obsessed with sharing gpus across workloads, albeit my interest has more dwealt in virtual desktop/vdi/cloud gaming. The people making the chips just want to charge sooo much money for sharing a gpu, seem to make it hard as they can.

Nice medium article on the challenges here too. https://grgalex.medium.com/gpu-virtualization-in-k8s-challen...

[deserialized]

Another for the pile https://github.com/cnvrg/metagpu

[grgalex]

Take a look at the Medium article [1] and it will be clear to you that this is not the same.

Each complete GPU sharing approach must have:

   - A mechanism to facilitate sharing (security, isolation, avoiding OOM errors).

   - A K8s integration.

Most approaches (like the one you mentioned above) lack a mechanism and simply work around the 1-1 GPU allocation on Kubernetes by advertizing more devices per physical GPU.

Those are not viable solutions.

Please take a look at Paragraph 5 ("The real challenge of GPU virtualization on K8s") onwards as well as the repo notes.

[1]: https://grgalex.medium.com/gpu-virtualization-in-k8s-challen...

[deserialized]

Thanks for the clarification, it took me rereading the article a couple more times to fully sink-in lol

Great write-up! I'm eager to test a few of these methods out in the lab

[grgalex]

Thanks, any feedback is welcome if you do try it out at some point :)