Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

GPUs != Neuron Devices (when it comes to instance selector) #8243

Merged
merged 1 commit into from
Feb 24, 2025

Conversation

dims
Copy link
Contributor

@dims dims commented Feb 22, 2025

ec2-instance-selector has a totally different option for Neuron Devices ( https://github.com/search?q=repo%3Aaws%2Famazon-ec2-instance-selector+accelerators&type=code ).

So the gpus field in instanceSelector does NOT cover Neuron Devices. For example if you say gpus: 0 you may still get an instance type with a neuron device/accelerator.

To explicitly specify the number of neuron devices you need to add another field neuron_devices for example see:

    instanceSelector:
      vCPUs: 8
      gpus: 0
      neuron_devices: 0
      memory: 16GiB
      cpuArchitecture: x86_64

There was a bug in ec2-instance-selector as well in the accelerator selection, which has been fixed and picked up in this PR

@dims dims force-pushed the gpus-are-not-accelerators branch from e796610 to 0567b13 Compare February 22, 2025 16:16
@dims
Copy link
Contributor Author

dims commented Feb 22, 2025

we also need aws/amazon-ec2-instance-selector#414

@dims
Copy link
Contributor Author

dims commented Feb 22, 2025

related to #7897

@dims dims force-pushed the gpus-are-not-accelerators branch from 0567b13 to f144ea5 Compare February 24, 2025 18:32
@dims dims marked this pull request as ready for review February 24, 2025 18:33
@dims dims enabled auto-merge (squash) February 24, 2025 18:33
@dims dims force-pushed the gpus-are-not-accelerators branch from f144ea5 to b6b4193 Compare February 24, 2025 18:35
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
@dims dims force-pushed the gpus-are-not-accelerators branch from b6b4193 to f96b8d7 Compare February 24, 2025 18:38
@dims dims changed the title GPUs != Accelerators GPUs != Neuron Devices Feb 24, 2025
@dims dims changed the title GPUs != Neuron Devices GPUs != Neuron Devices (when it comes to instance selector) Feb 24, 2025
@dims dims merged commit 2eee4c4 into eksctl-io:main Feb 24, 2025
13 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants