Replies: 2 comments
-
Being worked on in #2930! |
Beta Was this translation helpful? Give feedback.
0 replies
-
This is now merged in #4365! Great thanks to @kristopolous. Closing this discussion now. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello SkyPilot,
I've just found your framework, but I'm using a vast.ai service. It's a service that allows you to rent GPUs so cheap no other cloud provider can beat them.
It's not a promo, but just take a look at their pricing: https://cloud.vast.ai/create/, especially at interruptible (spot). You can get an H100 rented for 0.5-1.2$/hour, if you find the right deal (just 0.5$!!!). This is, of course, spot. Gcloud hosts the same for 4.2 - almost 4-8 time more expensive.
It's so cheap because it's non-professional renting, for example some university may rent their GPU installation over a weekend and still be profitable. There are hundreds of RTX4090 rigs too. 4090 rigs go around 0.1-0.15$/hour.
Many users would want such a cheap pricing; and I'll probably start migrating with skypilot because I want to a) simplify my spot instance work for future inference and b) use Google TPUs because they have great prices too.
Could you please consider implementing a vast.ai interface into the system?
Cheers.
(I'm not selling, it's just a truly great service)
Beta Was this translation helpful? Give feedback.
All reactions