GPU Costs and Availability
Discussions center on the high costs of GPUs, comparisons between cloud providers like AWS and alternatives such as CoreWeave or Runpod, and debates over renting versus buying hardware for compute-intensive tasks like AI training and inference.
Activity Over Time
Top Contributors
Keywords
Sample Comments
Looking at GPU cost and availability, I agree with you.
I guess the cost model doesn't work because you're buying gpu that you use about 0.1% of the day
Cloud GPU costs are way higher than actual GPU costs though.
How much are you paying to keep this up? Is it expensive to run GPU instances like this?
Only if they were buying GPU capacity.
Servers with GPU's are not going to be cheap :)
Running costs are about 30% of what an equivalent AWS GPU setup would be
Conisder adding CoreWeave for GPUs. The A100 there is about half the cost of V100 on large clouds.
I doubt it is a priority. But I can certainly recommend Amazon GPU-enabled instances (~0.6$/hour/GPU, not that much actually).
The top competitors are very likely using GPUs, I guess that would make this cost much much lower...