cluster_sparse_solver iparm(2)

cluster_sparse_solver iparm(2)

Hello,

How is the feature iparm(2) = 10 supposed to work with cluster_sparse_solver? If I adapt cl_solver_unsym_distr_c.c by changing the value of iparm[ 1] to 10, with 2 processes, it hangs on both Linux with IntelMPI and macOS with MPICH, and it segfaults with more than 2 processes.

Moreover, there is some unwanted output:

$ mpirun -np 2 ./a.out
RANK # 0 Total mem 0
RANK # 1 Total mem 0

<deadlock>

$ mpirun -np 4 ./a.out

===================================================================================
=   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
=   PID 48932 RUNNING AT 
=   EXIT CODE: 11
=   CLEANING UP REMAINING PROCESSES
=   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
===================================================================================
YOUR APPLICATION TERMINATED WITH THE EXIT STRING: Segmentation fault: 11 (signal 11)

Could this be fixed please?

Thanks.

 

AttachmentSize
Downloadtext/x-csrc cl_solver_unsym_distr_c.c11.87 KB
7 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

Have you checked the problem with the latest MKl 2018 u1 ( released at Nov '17)?

I've checked this problem with Intel(R) Math Kernel Library Version 2018.0.1 Product Build 20171007 for Intel(R) 64 architecture applications on all OSes.

With the Fortran example, and iparm(2) = 10, it also segfaults with more than 2 processes, but it seems OK with 2 processes, though there is a lot of unwanted output.

$ mpirun -np 4 _results/intel_intelmpi_lp64_intel64_a/cl_solver_sym_distr_f.exe 
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source             
cl_solver_sym_dis  000000000652FC3D  Unknown               Unknown  Unknown
libpthread-2.26.s  00007FC1AB87F160  Unknown               Unknown  Unknown

$ mpirun -np 2 _results/intel_intelmpi_lp64_intel64_a/cl_solver_sym_distr_f.exe
RANK # 0 Total mem 0
RANK # 1 Total mem 0
Memory allocated on phase  11 on Rank # 0       0.0000 Gb
Memory allocated on phase  11 on Rank # 1       0.0000 Gb

 

 

Attachments: 

AttachmentSize
Downloadapplication/octet-stream cl_solver_sym_distr_f.f7.49 KB

Can you reproduce these errors on both Linux and macOS with both C and Fortran?

Thanks.

yes, we see the problem on our side.  We will record it and fix it in future release. the next time, If any new issue, please feel free to submit to Online Service Center

Hello,

we fixed the problem and if you have access to the  Online Service Center, then you may submit ticket there and I will give you the engineering build to check if the problem exists on your side. 

thanks,

Gennady

Leave a Comment

Please sign in to add a comment. Not a member? Join today