topic Thanks mecej4! in Intel® oneAPI Math Kernel Library

problem when using cluster solver with distributed CSR format of input data

kybsh2006 — Sun, 04 Jun 2017 05:49:49 GMT

The following error encountered when using cluster solver:

ERROR during symbolic factorization: -2

The example (cl_solver_sym_distr_f.f) provided with Cluster compiler works well. But the matrix I provided in the attachment produces the above error.

I used Intel MPI to compile the code (also in the attachment). The executable file compiled through IntelMPI yields the above error. But it can run using OpenMPI's mpirun.

Could you help me check my code? Both the source code and data are in the attachment.

Thank you very much!

Qian

I have not gone through your

mecej4 — Sun, 04 Jun 2017 11:14:00 GMT

I have not gone through your code in detail, but something that I saw made me suspicious: there are three instances of '36' in the source files, suggesting that the number of equations to be solved is 36. This seems to conflict with the two data files, which seem to contain data for a set of 24 equations.

It may be easier to work out the bugs in a test program using a smaller number of equations on a single node, before switching to an MPI version and a large problem size.

Thanks mecej4!

kybsh2006 — Sun, 04 Jun 2017 17:36:32 GMT

Thanks mecej4!

The sparse matrix is distributed on two processors. Each processor has 24 equations. The number of equations for the final assembled sparse matrix is 36.

The code works on one processor to solve the 24 equations, which is just a part the final sparse matrix.

The input data is similar to the example provided by Intel compiler (cl_solver_sym_distr_f.f).

Qian

The problem has been solved.

kybsh2006 — Sun, 04 Jun 2017 23:41:29 GMT

The problem has been solved. It is my sparse matrix that causes the problem.

The correct sparse matrix is attached, which generates the correct solution.

Hi Kybsh2006,

Ying_H_Intel — Mon, 05 Jun 2017 01:10:49 GMT

Hi Kybsh2006,

Thank you a lot for the sharing the correct solution

Best Regards,

Ying