-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Much slower than torch.linalg.solve
#16
Comments
Hi @uranium11010 , Sparse solvers are tricky. There are lots of sparse solvers out there (see for example SuiteSparse). Each of them have different use cases and are ideal for different purposes. In this case,
in any case choosing the right algorithm for a sparse solve is important. I've been thinking of adding additional backends (LU, UMFPACK) to this package but haven't gotten to it yet. Maybe in the future. |
Ah I see--that makes sense to me. Thanks a lot and looking forward to seeing the additional backends :) |
I wrote the following code to benchmark against
torch.linalg.solve
for large sparse matrices. Here,A
is a 5000x5000 coalesced COO matrix with 10 nonzero elements in every row.b
is a length-5000 vector.The output:
I was hoping to use
torch_sparse_solve
to solve some large systems quickly and in a memory-efficient manner, so it would be great if this performance issue can be looked into. Thanks!The text was updated successfully, but these errors were encountered: