search

Lightning-AI / lightning-thunder

Make PyTorch models up to 40% faster! Thunder is a source to source compiler for PyTorch. It enables using different hardware executors at once; across one or thousands of GPUs.
Apache License 2.0
1.09k stars 63 forks source link

Add the torch.compile executor as a test executor #299

Open carmocca opened 2 months ago

carmocca commented 2 months ago

🚀 Feature

Motivation

Our op coverage for the torch.compile executor could be better, currently we only add it in the partial from in https://github.com/Lightning-AI/lightning-thunder/blob/3cd157b9bec6a450115f01125ee2653227988507/thunder/tests/framework.py#L337

Pitch

Add the TorchCompileCatExecutor and TorchCompileExecutor test executors to https://github.com/Lightning-AI/lightning-thunder/blob/3cd157b9bec6a450115f01125ee2653227988507/thunder/tests/framework.py#L242 and work through the errors that pop up in CI

Additional context

Requires #140 to land first.

cc @apaz-cli

mruberry commented 2 months ago

triage review — what would be the increase in e2e test time for add this? Should we consider some additional mechanisms of reducing CI time before increasing it?