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

add al2keplergpu build recipe to build gpu amis for kepler arch #153

Merged
merged 1 commit into from
Oct 18, 2023

Conversation

prateekchaudhry
Copy link
Contributor

@prateekchaudhry prateekchaudhry commented Oct 17, 2023

Summary

With the release of 20230912 AMI, default Nvidia and Cuda drivers were updated in AL repos. These further later added support for P5 instances but lost support for P2 instances (Kepler architecture), for which Nvidia 470 is the latest supported driver.

This PR adds a new al2keplergpu build recipe to support building AL2 ECS Optimized GPU AMIs with this package for instances with Kepler arch GPUs (such as P2 type). This recipe is NOT a new supported ECS platform, no new AMI types are published for this type. Rather it is a build recipe for anyone to build their own ECS AMIs for Kepler instances with GPUs on Kepler architecture.

This will enable customers to create their own GPU supported AMIs for Kepler based GPU instances (such as P2.xlarge) by

  • Setting the required Agent and source AL2 versions in release.auto.pkrvars.hcl or overrides.auto.pkrvars.hcl
  • Then running REGION=<region> make al2keplergpu

Implementation details

  • Adds new packer file for al2keplergpu and new makefile target
  • Changes in scripts/enable-ecs-agent-gpu-support.sh to pin package versions to supported versions for Kepler type GPUs. Also, exclude nvidia and cuda packages in the built AMI from yum updates to keep Kepler support for the built ami while maintaining regular updates.

Testing

Make amis REGION=us-west-2 make al2keplergpu and REGION=us-west-2 make al2gpu

  • Build al2keplergpu ami, run functional test suite on p2.xlarge instance, verify tests succeed.
  • Verified Nvidia driver 470.182.03 and Cuda 11.4 is installed.
  • Verified instance is registered with ecs.capability.gpu-driver-version:470.182.03
  • Verified running yum update -y does not update Nvidia/Cuda drivers. /etc/yum.conf has the exclusion exclude=*nvidia* *cuda*
  • Build al2gpu ami, run functional test suite on g4dn.xlarge instance, verify tests succeed.
  • Verify Nvidia driver 535.54.03 and Cuda 12.2 is installed.
  • Verified instance is registered with ecs.capability.gpu-driver-version:535.54.03

New tests cover the changes:
No new tests

Description for the changelog

add al2keplergpu build recipe to build gpu amis for kepler arch

Licensing

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

@prateekchaudhry prateekchaudhry force-pushed the p2support branch 2 times, most recently from 68516ea to 46ca787 Compare October 18, 2023 00:02
@prateekchaudhry prateekchaudhry changed the title add al2p2gpu build recipe to build gpu amis for p2 instances add al2keplergpu build recipe to build gpu amis for kepler arch Oct 18, 2023
@prateekchaudhry prateekchaudhry force-pushed the p2support branch 2 times, most recently from 5cd4798 to 69eb053 Compare October 18, 2023 00:20
@prateekchaudhry prateekchaudhry merged commit ee97bc4 into aws:main Oct 18, 2023
@danehlim danehlim mentioned this pull request Oct 23, 2023
@singholt singholt mentioned this pull request Oct 24, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants