Functions used for testing

Benchmarks for similar workloads among the cloud providers

Hello GPU

Hello GPU is a basic test with no dependencies. It simply boots a machine with a GPU and returns a string.

Hello Torch

Hello Torch is a basic test with Torch as a dependency. It loads Torch and runs some simple matrix multiplications on the GPU.

Hello TTS

Hello TTS is a test that runs the bark text-to-speech model on a single sentence. This test is small, but it's for a real use case that includes loading a big model into memory and running inference.

About

See the code used at github.com/viktorfa/serverless-gpu-samples