Tridiagonal solver in oneMKL for DPC++

[Oluwatosin]

Hello, is there are tridiagonal solver routine in oneMKL for data parallel c++ similar to dgtsv() routine in c/c++ intel mkl? Thank you.

[Gennady_F_Intel · Accepted solution]

yes, please check the API description here: https://software.intel.com/content/www/us/en/develop/documentation/oneapi-mkl-dpcpp-developer-reference/top/lapack-routines/getrs.html

[Gennady_F_Intel]

is this what are you looking for?

[Oluwatosin]

Thanks for replying, I will work with the routine.

[Oluwatosin]

One more question, I would be comparing results of this routine with an algorithm based on cyclic reduction, it has been designed to execute with a size specified at compilation time.  Maximum size is about 90,000,000 . Do I have to specify a memory size when requesting for a compute node (example- qsub -I -l nodes=1:gpu:ppn=2 -l mem=32G -d .)  in order to execute the code with this size?

[Gennady_F_Intel]

I think yes, You have to estimate the memory requested by getrs for this case and specify this size when will make the queue submission.

-Gennady

[Oluwatosin]

Alright, thank you!

[Gennady_F_Intel]

The issue is closing and we will no longer respond to this thread. If you require additional assistance from Intel, please start a new thread. Any further interaction in this thread will be considered community only.