[MITgcm-support] Error: No. of processes not equal to nPx*nPy

Yuan Lian lian at email.arizona.edu
Thu Apr 17 00:14:26 EDT 2008


Try to run the code with sNx=90, sNy=20, nPx=1 and nPy=2. This might help.
I had problems to run codes when I used multiple processors in x direction.

Yuan

Marcelo Barreiro wrote:
> Hello,
>
> I just downloaded the code using CVS, and I am trying to run
> verification/global_ocean.90x40x15 experiment using MPI (MPICH). 
>
> 1. I built an opt file for my machine with the libraries and compilers needed. 
> 2. In /build I did ../../../tools/genmake2 -mods=../code -mpi
> -of=../../../tools/build_options/opt_chpc
> 3. make depend
> 4. make 
> 5. In input ./prepare.run
> 5. ln -s ../input/* .
> 6. Used Loadleveler to submit: llsubmit llsubmit.test 
>
> Everything goes well if I use one processor and the program runs. However, if
> I change the SIZE.h to try to use 2 processors (mpirun -np 2 ...) the program
> compiles just fine but won't run and gives the error message:
>
> S/R EEBOOT_MINIMAL: No. of processes not equal to nPx*nPy     1     2
> S/R EEBOOT_MINIMAL: No. of processes not equal to nPx*nPy     1     2
> ABNORMAL END: PROGRAM MAIN
> ABNORMAL END: PROGRAM MAIN
>
> The SIZE.h file I am using has
>
>      &           sNx =  45,
>      &           sNy =  40,
>      &           OLx =   2,
>      &           OLy =   2,
>      &           nSx =   1,
>      &           nSy =   1,
>      &           nPx =   2,
>      &           nPy =   1,
>      &           Nx  = sNx*nSx*nPx,
>      &           Ny  = sNy*nSy*nPy,
>      &           Nr  =  15)
>
> and eedata
>  &EEPARMS
>  nTx=1,
>  nTy=1,
>  usingMPI=.TRUE.,
>  &
>
> As far as I understand I am satisfying the condition nPx*nPy=np that is needed
> to run the code. But what am I missing? 
>
> Thanks for your help!
>
> Marcelo
>
>
>  
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
>
>
>   




More information about the MITgcm-support mailing list