Can you add/remove GPUs to a Dedicated CPU Linode after creation?
I am thinking of the following use case: using a Dedicated CPU Linode as a remote development environment, and then using GPUs to accelerate ML tasks when necessary. With AWS and GCP you are able to add/remove GPUs as needed so that you aren't paying for a GPU the entire time if you only use it say 10% of the time.
Is this possible with Linode? Can I add/remove individual GPU/s to a pre-existing Dedicate CPU Linode as needed? Or would I have to create a separate new Dedicated GPU Linode, transfer over to that instance to execute, and then delete the instance from my account once the GPU-accelerated computation is completed?
Thanks in advance.
2 Replies
Hello,
So far as I know, you would need to create a GPU plan to utalize a GPU, I don't believe they can be added or removed from a dedicated plan. Also, the resource allocations are quite different between GPU and dedicated Linode's.
Hi @hari_seldon -
You can use our resize feature to upgrade to a GPU Linode instance from your existing Dedicated Linode. Once you're finished using the GPU, you can downsize the Linode back to a Dedicated plan. Here's our guide on resizing a Linode:
We also have a post on the Community Questions site that provides some considerations and steps on downsizing a Linode.
I hope this information helps!