Skip to content
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鈥檒l occasionally send you account related emails.

Already on GitHub? Sign in to your account

CSR: Relaxing constraints to s_addmm_out_sparse_dense_cuda_worker #59099

Open
Tracked by #60854
aocsa opened this issue May 27, 2021 · 0 comments
Open
Tracked by #60854

CSR: Relaxing constraints to s_addmm_out_sparse_dense_cuda_worker #59099

aocsa opened this issue May 27, 2021 · 0 comments
Assignees
Labels
module: sparse Related to torch.sparse open source triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module

Comments

@aocsa
Copy link
Contributor

aocsa commented May 27, 2021

馃殌 Feature

Given that cusparse APIs support setting ldb (leading dimension of B).

if(dense.stride(0) == 1 && dense.stride(1) == dense.size(0)) {

The previous condition is too strict condition as dense.stride(1) can be larger than dense.size(0). The first stride stride(1) can be size(0) * s where s is some positive integer.

Additional context

This function is used by addmm_out_sparse_csr_dense_cuda (CSR) and s_addmm_out_sparse_dense_cuda_worker (COO) tensor's operation.

Mentioned here:
#59012 (comment)

cc @aocsa @nikitaved @pearu @cpuhrsch @IvanYashchuk @mruberry

@aocsa aocsa added module: sparse Related to torch.sparse open source labels May 27, 2021
@aocsa aocsa added this to To do in Sparse tensors via automation May 27, 2021
@ngimel ngimel added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label Jun 1, 2021
@IvanYashchuk IvanYashchuk self-assigned this Jul 6, 2021
@pearu pearu moved this from To do to To Do: CSR in Sparse tensors Aug 10, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
module: sparse Related to torch.sparse open source triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Projects
Sparse tensors
To Do: CSR
Development

No branches or pull requests

3 participants