[mpich-discuss] MPI profiling help

Lu, Huiwei huiweilu at mcs.anl.gov
Mon Aug 4 11:28:19 CDT 2014


I have tried MPICH 3.1.2 on our BG/Q system. It was built fine.

Here is the log:
$ which powerpc64-bgq-linux-gcc
/bgsys/drivers/V1R2M1/ppc64/gnu-linux/bin/powerpc64-bgq-linux-gcc
$ ./configure --host=powerpc64-bgq-linux --with-device=pamid:BGQ --with-file-system=bg+bglockless
$ grep '^CC=' config.log
CC='powerpc64-bgq-linux-gcc'
$ make -j 32

Arnamoy: can you try again with ‘pamid:BGQ’? I am using the options given in mpich-3.1.2/README, which said '--with-device=pamid:BGQ’ instead of '--with-device=pamid’.

—
Huiwei Lu
Postdoc Appointee
Mathematics and Computer Science Division, Argonne National Laboratory
http://www.mcs.anl.gov/~huiweilu/

On Aug 4, 2014, at 10:26 AM, Antonio J. Peña <apenya at mcs.anl.gov> wrote:

> 
> Hi,
> 
> For the MPICH 3.1.2, that's a known issue of the IBM compiler that we have recently worked around. Please try to pull the most recent source from our repository.
> 
> For the others, at a glance I have no clue of what may be wrong there. We will need your full configure log and make errors. Anyway, I'm CC'ing the discuss mailing list, where other folks with more experience building MPICH in BG/Q may give you a hint.
> 
> 
> On 08/04/2014 10:20 AM, Arnamoy Bhattacharyya wrote:
>> Hi Antonio;
>> 
>> Can I bug you again a bit?  It's not on MPI profiling but how to install and use different versions of MPICH in BG/Q.  So I have followed the instructions from this webpage (https://wiki.mpich.org/mpich/index.php/BGQ#override_configure_settings_with_the_compiler_environment_variables),  and I build with the first method (building with gnu toolchain)
>> 
>> I do:
>> 
>> 1.  export PATH=$PATH:/bgsys/drivers/V1R2M1/ppc64/gnu-linux/bin
>> 2.  ./configure --prefix=<directory> --host=powerpc64-bgq-linux --with-device=pamid
>> 3.  make
>> 4.  make install
>> 
>> I have tried this method for MPICH-1.5, MPICH 3.0.1, MPICH 3.1.2, but not a single succeeded in "make" (the errors I have given below).
>> 
>> Am I missing something in the building process?  Can you tell me what I am doing wrong?
>> 
>> Thanks again;
>> 
>> ERRORS:
>> 
>> MPI 1.5 :
>> 
>> /home/arnamoy/summer14/mpi-1.5/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h: In function 'MPIDI_Recv':
>> /home/arnamoy/summer14/mpi-1.5/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h:120: error: 'struct MPIDI_Request' has no member named 'userbuf'
>> /home/arnamoy/summer14/mpi-1.5/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h:121: error: 'struct MPIDI_Request' has no member named 'userbufcount'
>> /home/arnamoy/summer14/mpi-1.5/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h:122: error: 'struct MPIDI_Request' has no member named 'datatype'
>> 
>> 
>> MPI 3.0.1 :
>> 
>> In file included from /home/arnamoy/summer14/mpi-3.0.1/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpid_irecv.h:26,
>>                  from /home/arnamoy/summer14/mpi-3.0.1/mpi-src/src/mpid/pamid/include/mpidpost.h:38,
>>                  from ./src/include/mpiimpl.h:3739,
>>                  from src/mpi/attr/attrutil.c:16:
>> /home/arnamoy/summer14/mpi-3.0.1/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h: In function 'MPIDI_Recv':
>> /home/arnamoy/summer14/mpi-3.0.1/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h:120: error: 'struct MPIDI_Request' has no member named 'userbuf'
>> /home/arnamoy/summer14/mpi-3.0.1/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h:121: error: 'struct MPIDI_Request' has no member named 'userbufcount'
>> /home/arnamoy/summer14/mpi-3.0.1/mpi-src/src/mpid/pamid/include/../src/pt2pt/mpidi_recv.h:122: error: 'struct MPIDI_Request' has no member named 'datatype'
>> 
>> MPI 3.1.2
>> 
>> "src/binding/fortran/mpif_h/setbotf.f", line 1.7: 1515-019 (S) Syntax is incorrect.
>> "src/binding/fortran/mpif_h/setbotf.f", line 1.26: 1515-019 (S) Syntax is incorrect.
>> "src/binding/fortran/mpif_h/setbotf.f", line 2.1: 1515-019 (S) Syntax is incorrect.
> 
> 
> -- 
> Antonio J. Peña
> Postdoctoral Appointee
> Mathematics and Computer Science Division
> Argonne National Laboratory
> 9700 South Cass Avenue, Bldg. 240, Of. 3148
> Argonne, IL 60439-4847
> 
> apenya at mcs.anl.gov
> www.mcs.anl.gov/~apenya
> _______________________________________________
> discuss mailing list     discuss at mpich.org
> To manage subscription options or unsubscribe:
> https://lists.mpich.org/mailman/listinfo/discuss




More information about the discuss mailing list