topic Re: Help on running GMRES with MKL in C++ settings to use ILP64 libraries in Intel® oneAPI Math Kernel Library

Help on running GMRES with MKL in C++ settings to use ILP64 libraries

houjy — Sat, 24 Aug 2024 06:10:20 GMT

I am new to MKL library. I tried to use to use it to solve a sparse system matrix using “GMRES” functions in MKL library.

my way using ILP64 is:

#define MKL_ILP64

and my problem is call dfgmres_init(), RCI_request=‭-3689348814741910324.

Thank you very much

Re: Help on running GMRES with MKL in C++ settings to use ILP64 libraries

Gajanan_Choudhary — Sat, 24 Aug 2024 16:13:26 GMT

Hi @houjy,

Thanks for reaching out. That is not the way it works. oneMKL has separate sub-libraries/interface layer built for 32-bit integers (libmkl_lp64) and 64-bit integers (libmkl_ilp64). When you are compiling your application, you have to link to one of them based on what integer size you plan to use with oneMKL. For 64-bit integers, you also pass in -DMKL_ILP64 in the compile line for your application objects.

Please use the Intel Link Line Advisor for getting the correct link line for your application. The "Select interface layer" field of in the link is where you select 32- or 64-bit integers.

Hope that helps!

Re: Help on running GMRES with MKL in C++ settings to use ILP64 libraries

houjy — Sun, 01 Sep 2024 12:46:55 GMT

Thank you very much. I have used ILP64 successfully.

But I have another question,When I use fgmres to solve large systems of linear equations, some values in the ipar array are equal to -5589219598436180572

Thank you very much