Skip to content

Conversation

fonnesbeck
Copy link
Member

First pass at using Numba with Pathfinder. It works, but is not yet performant (order of magnitude slower at the moment).

@ricardoV94
Copy link
Member

You're implementing way too much stuff. All the contiguous matmul stuff... why? Numba is capable of doing that much better and will dispatch to blas... which you want to do.

What was the old pathfinder doing and why can't you just compile PyTensor stuff to numba?

@fonnesbeck
Copy link
Member Author

OK, I figured we wanted to use PyTensor where possible. Back to the drawing board!

@ricardoV94
Copy link
Member

OK, I figured we wanted to use PyTensor where possible. Back to the drawing board!

Yes we do, but you shouldn't have to implement any new Ops, like you were doing?

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.

2 participants