Segmentation fault with MKL PBLAS

mcastellana — Sun, 08 Apr 2012 03:12:31 GMT

Hello, I am trying to compile and run a sample code for a MKL PBLAS routine. I found the code herehttp://software.intel.com/en-us/articles/using-cluster-mkl-pblasscalapack-fortran-routine-in-your-c-program/ . You can find the code herepdgemv.c

I compile the code on a machine with Intel Mkl installed with the following command

mpicc -w -o pdgemv.o pdgemv.c -I/opt/intel/mkl/10.2.4.032/include -L/opt/intel/mkl/10.2.4.032/lib/em64t -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64 -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -liomp5 -lpthread

and get the warning

/opt/intel/Compiler/11.1/075/lib/intel64/libimf.so: warning: warning: feupdateenv is not implemented and will always fail

Then when I try to run I get segmentation fault:

[michelec@della3 pdgemv]$ mpirun -n 2 ./pdgemvlibrdmacm: couldn't read ABI version.

librdmacm: assuming: 4

CMA: unable to get RDMA device list

--------------------------------------------------------------------------

[[15168,1],0]: A high-performance Open MPI point-to-point messaging module

was unable to find any relevant network interfaces:

Module: OpenFabrics (openib)

Host: della3

Another transport will be used instead, although this may result in

lower performance.

--------------------------------------------------------------------------

librdmacm: couldn't read ABI version.

librdmacm: assuming: 4

CMA: unable to get RDMA device list

[della3:03459] *** Process received signal ***

[della3:03459] Signal: Segmentation fault (11)

[della3:03459] Signal code: Address not mapped (1)

[della3:03459] Failing at address: 0x44000098

[della3:03458] *** Process received signal ***

[della3:03458] Signal: Segmentation fault (11)

[della3:03458] Signal code: Address not mapped (1)

[della3:03458] Failing at address: 0x44000098

[della3:03459] [ 0] /lib64/libpthread.so.0 [0x353900eb70]

[della3:03459] [ 1] /usr/local/openmpi/1.4.3/intel111/x86_64/lib64/libmpi.so.0(MPI_Comm_size+0x60) [0x2b821ceea410]

[della3:03459] [ 2] /opt/intel/Compiler/11.1/075/mkl/lib/em64t/libmkl_blacs_intelmpi_lp64.so(Cblacs_pinfo+0x93) [0x2b821aca1293]

[della3:03459] *** End of error message ***

[della3:03458] [ 0] /lib64/libpthread.so.0 [0x353900eb70]

[della3:03458] [ 1] /usr/local/openmpi/1.4.3/intel111/x86_64/lib64/libmpi.so.0(MPI_Comm_size+0x60) [0x2b206a7b6410]

[della3:03458] [ 2] /opt/intel/Compiler/11.1/075/mkl/lib/em64t/libmkl_blacs_intelmpi_lp64.so(Cblacs_pinfo+0x93) [0x2b206856d293]

[della3:03458] *** End of error message ***

--------------------------------------------------------------------------

mpirun noticed that process rank 1 with PID 3459 on node della3 exited on signal 11 (Segmentation fault).

--------------------------------------------------------------------------

[della3:03457] 1 more process has sent help message help-mpi-btl-base.txt / btl:no-nics

[della3:03457] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages

Could you please help me?

Thanks!

Michele

Segmentation fault with MKL PBLAS

Gennady_F_Intel — Sun, 08 Apr 2012 03:51:25 GMT

we have never met with this issue before - please see here may be this recomendation helps you.

Segmentation fault with MKL PBLAS

mcastellana — Sun, 08 Apr 2012 19:24:28 GMT

Hello Gennady,

Thank you for your reply. Unfortunately the link didn't help: it seems to me that I have a different problem on running time. The segmentation fault occurs when "Cblacs_pinfo( &myrank_mpi, &nprocs_mpi ) ;" is called. Do you have any idea?

Thank you!

Michele

Same problem and I cant

Jehandad_K_ — Thu, 14 Aug 2014 04:06:42 GMT

Same problem and I cant figure it out !

I found the problem, I was

Jehandad_K_ — Thu, 14 Aug 2014 05:02:34 GMT

I found the problem, I was linking to intel mpi , when I linked to openmpi mkl the problem is solved !

topic Segmentation fault with MKL PBLAS in Intel® oneAPI Math Kernel Library

Segmentation fault with MKL PBLAS

Segmentation fault with MKL PBLAS

Segmentation fault with MKL PBLAS

Same problem and I cant

I found the problem, I was