question about the performance of oneapi

[Yelim]

Hi, I have a question about the performance of oneapi.
I'm evaluating several solvers to use a matrix analysis engine for my solution.
As a result of the evaluation, suitesparse's cholmod is faster than your engine. I want to use oneapi because I want to use your schur function as well.

When analysis - factoriztion - solving is executed, pardiso is 10 times slower than cholmod. Is there any way to speed it up even more?
Feel free to reply if there is anything I need to provide more information on.

thank you.

* matrix info
n:8833
nnz:3256859
matrix file : 0.0.facto.ijv

[ShanmukhS_Intel]

Hi Yelim,

Thanks for posting in Intel Communities.

Could you please get back to us with more details like (OS environment details, MKL version, etc)?  We would like to request you share the input file in a readable format as we could not be able to access the shared .ijv file.

If you are using an older version of oneAPI, we suggest you try working on the latest oneAPI and let us know if issue persists.

Best Regards,

Shanmukh.SS

[Yelim]

OS : RHEL 7.9

MKL Version : Intel One API 2023

for ( i = 0; i < 64; i++ )
{
iparm[i] = 0;
}
iparm[0] = 1; /* No solver default */
iparm[1] = 3; /* Fill-in reordering from METIS */
iparm[3] = 0; /* No iterative-direct algorithm */
iparm[4] = 0; /* No user fill-in reducing permutation */
iparm[5] = 0; /* Write solution into x */
iparm[6] = 0; /* Not in use */
iparm[7] = 2; /* Max numbers of iterative refinement steps */
iparm[8] = 0; /* Not in use */
iparm[9] = 13; /* Perturb the pivot elements with 1E-13 */
iparm[10] = 0; /* Use nonsymmetric permutation and scaling MPS */
iparm[11] = 0; /* Not in use */
iparm[12] = 0; /* Maximum weighted matching algorithm is switched-off (default for symmetric). Try iparm[12] = 1 in case of inappropriate accuracy */
iparm[13] = 0; /* Output: Number of perturbed pivots */
iparm[14] = 0; /* Not in use */
iparm[15] = 0; /* Not in use */
iparm[16] = 0; /* Not in use */
iparm[17] = -1; /* Output: Number of nonzeros in the factor LU */
iparm[18] = -1; /* Output: Mflops for LU factorization */
iparm[19] = 0; /* Output: Numbers of CG Iterations */
iparm[26] = 1;
maxfct = 1; /* Maximum number of numerical factorizations. */
mnum = 1; /* Which factorization to use. */
msglvl = 1; /* Print statistical information in file */
error = 0; /* Initialize error flag */

I can't upload the project file because of my company regulation. sorry

and could you try to open ijv file using text editor or text mode?

[ShanmukhS_Intel]

Hi Yelim,

Thanks for sharing the details. Could you please share with us the performance statistics which were observed between suitesparse and oneAPI? We would require a minimal isolated code for further analysis.

Best Regards,

Shanmukh.SS

[Yelim]

Hi, Below is the data I evaluated

solver type	# thread (Host memory  : 3TB, OS : RHEL7.9)	reordering time[Hr]	factorization time[Hr]	sum(re. + fac.) time[Hr]	solve time[Hr]	assume 2000 step time[Hr]
mkl	32	0.00022	0.00008	0.00030	0.00001	0.023
suitesparse	32	0.00005	0.00014	0.00019	0.00000	0.007

thanks for your help

[ShanmukhS_Intel]

Hi Yelim,

Thanks for sharing the details.

As mentioned earlier, We would like to inform you that, for community support, we would require you to submit a minimum reproduction sample code specific to the performance issue that provides us the most relevant background information for triage. In case you require privacy, and are unable to share the issue / sample with us publicly and If you are a licensed oneAPI product customer and/or member of Intel’s oneAPI Academic Program please submit a ticket for Priority support so that your application can be handled with the required data protection and privacy regulations.

Best Regards,

Shanmukh.SS

[ShanmukhS_Intel]

Hi Yelim,

A gentle reminder:

Could you please get back to us with the earlier-mentioned details? It helps us recreate the issue under our environment and help you further.

Best Regards,

Shanmukh.SS