[mpich-discuss] Running problem with MPICH2

[Gus Correa]

Hi Basak

The cpi.c program in the examples directory in the MPICH2
source code tree is a simple test of MPICH2 functionality.
You can try compiling it with mpicc and running it with mpiexec
(or mpiexec.hydra).

It also tells you what is the typical/basic structure of an
MPI program, starting with MPI_Init, ending with MPI_Finalize,
and doing explicit communication among the participating processes
in between, via a variety of MPI function calls.
My recollection is that cpi.c only uses MPI_Reduce, but there are many
others MPI functions, of which you can find examples in
the references below.

For more information about how to write MPI programs, try the
LLNL tutorials:

https://computing.llnl.gov/tutorials/mpi/

or Peter Pacheco's book:
http://www.cs.usfca.edu/~peter/ppmpi/

or W. Gropp et al. books (Rajeev is a co-author in the 2nd book):

http://www.mcs.anl.gov/research/projects/mpi/usingmpi/
http://www.mcs.anl.gov/research/projects/mpi/usingmpi2/

I hope this helps,
Gus Correa

Chandra Bhanu Basak wrote:
> Dear Chan
> 
> Sorry to bother you again and excuse me for my ignorance -
> 
> does that means the same code (count.c), if I compile as -
> 
> mpicc count.c
> 
> will be compiled in MPI version? and I don't have to change the source 
> code??
> 
> Basak
> 
> 
> 
> On 27 May 2011 20:07, Anthony Chan <chan at mcs.anl.gov 
> <mailto:chan at mcs.anl.gov>> wrote:
> 
> 
>     count.c is a serial program, so it won't use MPI's multiprocessing
>     support
>     You need to make count.c as MPI program. Recent MPICH2
>     uses hydra as default process manager, so you don't need to use
>     mpd or mpdboot,...  See
> 
>     http://wiki.mcs.anl.gov/mpich2/index.php/Using_the_Hydra_Process_Manager
> 
>     A.Chan
> 
>     ----- Original Message -----
>      > Hi,
>      >
>      > I have tried to installed recent version of MPICH2 in Scientific
>     Linux
>      > 6.
>      > The machine contains 2 Xeon X5650 processors (each having 6 cores).
>      > Apparently I was able to install MPICH2 without any problem (thanks
>      > for the
>      > wonderful installation guide). "which mpiexec" and "which mpicc" show
>      > the
>      > correct PATH.
>      >
>      > mpiexec -n 2 ./count.out
>      >
>      > while running this homemade small counting program (count.c attached
>      > here),
>      > system monitor shows only one core (not even one processor) is taking
>      > the
>      > entire load. count.c was compiled using gcc (not mpigcc).
>      >
>      > Also, I found that the commands like mpd or mpdboot or mpdtrace
>      > returns
>      > "Command not found".
>      >
>      > I am attaching here the c.txt, m.txt (as advised in the guides) also
>      > the
>      > small counting program (count.c).
>      >
>      > It will be of great help if anyone can show me the root of the
>     problem
>      > and
>      > perhaps the solution.
>      >
>      > Regards,
>      >
>      > C. B. Basak
>      >
>      > _______________________________________________
>      > mpich-discuss mailing list
>      > mpich-discuss at mcs.anl.gov <mailto:mpich-discuss at mcs.anl.gov>
>      > https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>     _______________________________________________
>     mpich-discuss mailing list
>     mpich-discuss at mcs.anl.gov <mailto:mpich-discuss at mcs.anl.gov>
>     https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss