Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- Intel Community
- Software
- Software Development SDKs and Libraries
- Intel® oneAPI Math Kernel Library
- Question about sparse matrix

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page

kim__seongik

Beginner

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-09-2020
11:36 PM

213 Views

Question about sparse matrix

1. Is there an appropriate subroutine in mkl with respect to matrix assembly?

I made and used by myself, but it takes too long.

I want to something like below expression in fortran

(example-matlab case) global_matrix(dof,dof) = global_matrix(dof,dof) + element_matrix

2. Is there any subroutine in mkl to create a sparse matrix format?

3. Which subroutine should be used for sparse matrix multiplication?

I want to something like below expression in fortran

(example -matlab case) A*x=b (A : sparse matrix, x: vector, b:vector)

4. Is there a subroutine that can do in fortran mkl how to delete certain rows and columns?

I want to something like below expression in fortran

(example -matlab case) matrix(dof,dof) = [];

5. I want to use scalapack of intel mkl, but can't I use it except cluster??

When I used scalapack's dlasrt2 on the desktop, error lnk 2019 occurred.

1 Solution

Gennady_F_Intel

Moderator

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-10-2020
02:40 AM

204 Views

1. there are no such routines

2. you can try to use conversion from/to dense - CSR, CSC, and others supported sparse formats. Please refer to use MKL Reference to see more details

3. please refer to the documentation follow with this link https://software.intel.com/content/www/us/en/develop/documentation/mkl-developer-reference-c/top/bla...

You may see a lot of m*v routines with different sparse formal, e.x - tmkl_?csrgemv

4. There is no such routine.

5. you can build and run on the SMP system as well.

how did you link the case? how did you run it?

Link Copied

5 Replies

Gennady_F_Intel

Moderator

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-10-2020
02:40 AM

205 Views

1. there are no such routines

2. you can try to use conversion from/to dense - CSR, CSC, and others supported sparse formats. Please refer to use MKL Reference to see more details

3. please refer to the documentation follow with this link https://software.intel.com/content/www/us/en/develop/documentation/mkl-developer-reference-c/top/bla...

You may see a lot of m*v routines with different sparse formal, e.x - tmkl_?csrgemv

4. There is no such routine.

5. you can build and run on the SMP system as well.

how did you link the case? how did you run it?

Gennady_F_Intel

Moderator

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-10-2020
02:42 AM

203 Views

kim__seongik

Beginner

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-12-2020
07:35 PM

191 Views

Thank you for your reply !!

In number 5 my setting is below figure::

Gennady_F_Intel

Moderator

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-13-2020
02:12 AM

184 Views

Please check the user's guide on how to use the linker

-mkl=cluster

to link with Intel® MKL cluster components (sequential) that use Intel MPI.

Gennady_F_Intel

Moderator

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-13-2020
02:14 AM

183 Views

I missed that you use windows:

the linking would look like as follows: /Qmkl=cluster

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

For more complete information about compiler optimizations, see our Optimization Notice.