-
Notifications
You must be signed in to change notification settings - Fork 218
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
[CUSPARSE] Implement a sparse GEMV for CuSparseMatrixCSC * CuSparseVector #2488
Conversation
|
In practice, we don't know the sparsity pattern of the output and thus for very rare cases we want to do that in-place. Maybe I should remove the high-level |
I'll leave that up to you to decide what's the most likely scenario here. Adding a docstring on the behavior and the alternative is probably a good idea though. |
The most likely scenario is that the result will be dense, so I decided to remove the high-level dispatch for this sparse |
bef7991
to
893960e
Compare
#2484
I didn't added new tests because the current ones should check if the result is correct or not:
https://github.com/JuliaGPU/CUDA.jl/blob/master/test/libraries/cusparse/interfaces.jl#L267-L282
Note that a sparse GEMV
CuSparseMatrixCSR * CuSparseVector
is also implemented.