Skip to content

Conversation

wsmoses
Copy link

@wsmoses wsmoses commented Oct 10, 2024

See #476

Obviously should be reformatted and cleaned up a bit.

@oscardssmith
Copy link
Member

How does performance compare?

@avik-pal
Copy link
Member

From our lab meeting discussion earlier this morning, we decided to have an Enzyme special case for SimpleNonlinearSolve at the very least (I will add that via #458).

We can see the overall performance benefits here (in the NonlinearSolve context) later, but for SimpleNonlinearSolve, I would argue we want to save every second we can, and if adding a minor maintenance burden is the way to go (especially with NonlinearSolveBase switching the default backend to Enzyme if loaded), then we should do it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants