topic Re:migrating cusparseXnnz in Intel® oneAPI Math Kernel Library

migrating cusparseXnnz

sycl-developer — Wed, 13 Sep 2023 15:33:27 GMT

The link is https://docs.nvidia.com/cuda/cusparse/index.html#cusparse-t-nnz

This function computes the number of nonzero elements per row or column and the total number of nonzero elements in a dense matrix. Thanks for considering them in the future releases of oneMKL.

Re:migrating cusparseXnnz

IntelSupport — Fri, 15 Sep 2023 07:31:31 GMT

Hi,

Thanks for posting in Intel communities.

Thanks for sharing your requirement details. We are looking into your issue internally. We will get back to you soon with an update.

Regards,

Jilani

Re: migrating cusparseXnnz

Gajanan_Choudhary — Fri, 15 Sep 2023 15:26:57 GMT

Hi @sycl-developer,

I'm not sure if you noticed that the cusparse<t>nnz function in the latest cuSPARSE version 12.2 is in their "Legacy APIs" section. Over the last few releases, we have noticed that cuSPARSE has been slowly deprecating and removing their legacy APIs (since around version 10.2 or 11.0, I believe, but I'm not exactly sure). Given this situation, I won't be surprised if they deprecate this API in a future release and maybe replace it with an alternative.

That said, our current plan is to add many APIs to oneMKL Sparse BLAS in the coming year(s) that will be alternatives similar to cuSPARSE's APIs.

Regards,

Gajanan

Re:migrating cusparseXnnz

IntelSupport — Fri, 22 Sep 2023 10:40:32 GMT

Hi,

A gentle reminder:

Has the information provided helped? Could you please let us know if you need any other details?

Best Regards,

Jilani Shaik

Re:migrating cusparseXnnz

IntelSupport — Tue, 03 Oct 2023 07:01:04 GMT

Hi,

A gentle reminder:

Has the information provided helped? Could you please let us know if you need any other details?

Best Regards,

Jilani Shaik