topic Re: MPI Hello World fails in Intel® MPI Library

MPI Hello World fails

kmccall882 — Thu, 10 Mar 2022 17:47:25 GMT

I'm using IntelMpi 2021.5.1 on RedHat 8.5 with an NFS file system. My simple Hello World program fails with many error messages. Any help would be appreciated.

Here is the program:

int main(int argc, char *argv[]) { int rank, world_size, error_codes[1]; char hostname[128]; MPI_Comm intercom; MPI_Info info; MPI_Init(&argc, &argv); MPI_Comm_rank(MPI_COMM_WORLD, &rank); MPI_Comm_size(MPI_COMM_WORLD, &world_size); gethostname(hostname, 127); std::cout << "Hello from process on " << hostname << std::endl; MPI_Finalize(); }

I sourced /opt/intel/oneapi/setvars.sh before building the executable, and

here is the run command:

$ export I_MPI_PIN_RESPECT_CPUSET=0; mpirun ./parent_simple

Here are the abridged error messages. I eliminated many repetitions:

[1646934040.094426] [rocci:306332:0] ib_verbs.h:84 UCX ERROR ibv_exp_query_device(mlx5_0) returned 95: Operation not supported

[1646934040.113276] [rocci:306320:0] select.c:434 UCX ERROR no active messages transport to <no debug data>: self/memory - Destination is unreachable, rdmacm/sockaddr - no am bcopy

Abort(1090703) on node 0 (rank 0 in comm 0): Fatal error in PMPI_Init: Other MPI error, error stack:
MPIR_Init_thread(143)........:
MPID_Init(1310)..............:
MPIDI_OFI_mpi_init_hook(1974): OFI get address vector map failed
[1646934040.113302] [rocci:306315:0] select.c:434 UCX ERROR no active messages transport to <no debug data>: self/memory - Destination is unreachable, rdmacm/sockaddr - no am bcopy

Re: MPI Hello World fails

VarshaS_Intel — Mon, 21 Mar 2022 11:05:40 GMT

Hi,

Thanks for posting in Intel Communities.

We are unable to reproduce your issue at our end. We tried with your sample reproducer code and we were able to get the expected results.

We followed the below steps using the latest Intel MPI 2021.5 on a Linux machine:

1. Please find the below command:

For Compiling, use the below command: mpiicc -o hello hello.cpp For Running the MPI program, use the below command: export I_MPI_PIN_RESPECT_CPUSET=0;mpirun -bootstrap ssh -n 1 -ppn 1 ./hello

Could you please provide us with the OS details and FI provider you are using?

Please run the below command for cluster checker and share with us the complete log file.

clck -f ./<nodefile> -F mpi_prereq_user

Please find the attached screenshot for the expected results.

Thanks & Regards,

Varsha

Re: MPI Hello World fails

Xiao_Z_Intel — Thu, 24 Mar 2022 06:25:12 GMT

Hi Kurt,

Have you run the cluster checker as Varsha suggested? Could you please also run the following items and share with us the detailed results including the complete log files?

share the output of ucx_info -d and fi_info -v
run the code by enabling debug options using I_MPI_DEBUG=10 and FI_LOG_LEVEL=debug
run the code with tcp as your OFI* provider (FI_PROVIDER=tcp ) and enable debug options

Thanks,

Xiao

Re:MPI Hello World fails

Xiao_Z_Intel — Mon, 02 May 2022 20:27:31 GMT

Hi Kurt,

We did not heard back from you for the additional information and will close 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.

Best,

Xiao