ROSA HCP needs support for Accelerated computing > Amazon EC2 G6 instances powered by NVIDIA L4 Tensor Core GPU launched in April 2024

[maulik-modi22]

Which service is this feature request for?
Red Hat OpenShift Service on AWS
https://aws.amazon.com/about-aws/whats-new/2024/04/general-availability-amazon-ec2-g6-instances/
https://aws.amazon.com/ec2/instance-types/g6/

What are you trying to do?
To run inferences with HighRes VIT model to detect cracks and damages in concrete, we require Multi GPU VM

Describe the solution you'd like
Would like to see EC2 G6 instances that are powered by NVIDIA L4 Tensor Core GPU(L4) as supported instance types under Accelerated computing
https://docs.openshift.com/rosa/rosa_architecture/rosa_policy_service_definition/rosa-hcp-instance-types.html should have

Describe alternatives you've considered
Other GPUs such as g5.12xlarge(A10) and p3.8xlarge(V100) are too much expensive and cheaper GPU such as g4dn.12xlarge(T4) do not meet performance requirement. On the other hand g6.*(L4) series offers sweet spot for price/performance.

Additional context

[maulik-modi22]

@arendej, @wgordon17 , Could you please assign appropriate labels and seek feedback with PM team?

[maulik-modi22]

@wgordon17, Could you please assign appropriate labels and seek feedback with PM team to certify g6.*(L4) GPU?

[maulik-modi22]

@arendej, We are into 2025 now, As there is a huge bump in the price between T4 and A10.

[maulik-modi22]

@arendej, Can we expect this in Q1 2025?