Ask HN: How to build GPU compute marketplace?

1 pointsmichaeloblak2y ago4 comments

Is it possible? Let's say Alice has 2 GPUs idle at the moment, Ben has 1 GPU, and Chris needs 3 GPUs for the next 12 hours. How to build such a system, and what problems there might occur? How to handle turning off one's machine? Does it even make sense to run training or inference on such (latency)?

4 comments

ftxbro2y ago

There is already some like this including vast dot ai.

michaeloblakOP2y ago

Yeah, but how does it work exactly? Dropbox is rsync with UI, Instagram is a blog with pictures, but this?

How to access one’s compute? Overriding GPU drivers?

deserialized2y ago

It's just a remote docker context on a machine with the nvidia driver and container runtime already installed that you rent by the hour.

brucethemoose22y ago

More task specific compute pools (like Stable Horde) make lots of sense where applicable.

j / k navigate · click thread line to collapse