ventz/whisper-openai-container

question about performance in Lambda

Closed this issue · 2 comments

AFAIK Lambda doesn't provide access to GPU resources - how do these workflows run? Is it slow?

Hi @jvsteiner - Thats correct, Lambda's do not give GPU, thats why I built an"hybrid" that can be deployed as a standalone container for EC2/Fargate/external providers.

As an aside - Modal's cloud does let you run Functions on GPU backed containers. Imo, Modal also makes it 10x easier to run lambda/container (mixed, or each) workloads.

As far as speed on AWS Lambda's - my use case seems to fit within the 15 min max execution that Lambda provides.

checking out modal now - looks like a gamechanger...