Closed JohnTigue closed 1 year ago
AWS Blog, 2022-12 seems still no GPUs on Serverless AWS so need to stick to some EC2 based service (maybe EKS, maybe raw EC2?):
Many of the services selected are serverless, which will offer many benefits. At the time of writing, Stable Diffusion requires a GPU for inference. Amazon Elastic Compute Cloud (Amazon EC2) was selected because it provides GPUs.
And this gets back to that ECR rigmarole I went through in 2022. There's some pretty simple, boilerplate aws-cli command sequence to build an image and publish it in a private registry on ECR. We already have a registry set up. See the closed issues in dev_tools:
Going with ECS over EKS because the former automates the "undifferentiated heavy lifting" of the latter, which requires a trained Kubernetes team which we do not have. See Amazon ECS vs Amazon EKS: making sense of AWS container services
There are multiple ways to deploy Docker containers on AWS. Which one should we use?