[mpich-discuss] mpe error

[Mandar Gurav]

Hi Peter!

just follow these steps...
1. Compile program
        mpicc myprogram.c -mpe=mpilog

2. Execute program
        mpiexec -np 4 ./a.out

3. Log file will be generated "a.out.clog2". To open the file through
jumpshot use command
        jumpshot a.out.clog2

4. It asks(In my case) for slog2 file. Just say yes to convert to
slog2. Use convert button at the bottom. Exit and run again with
slog2...
        jumpshot a.out.slog2


--Mandar Gurav

On Tue, May 17, 2011 at 6:52 PM, Anthony Chan <chan at mcs.anl.gov> wrote:
>
> When you link with -mpe=mpilog, you don't need -lmpe flag.
> MPE generates trace file for each process and merges them during
> MPI_Finalize.  Can you post your program here if it isn't too big?
>
> A.Chan
>
> ----- Original Message -----
>> Hi,
>>
>> Thanks for the replies! To clarify, my program seems to work when I
>> compile and run the program without MPE.
>> But as soon as I add the "-mpe=mpilog" and the "-lmpe" flag, I get the
>> error message.
>> What does "merging" mean?
>>
>> Thanks!
>>
>> Peter
>>
>> >  Backtrace shows that merging failed. Would you be able to run
>> >  the same program without MPE ?
>> >
>> >  A.Chan
>> >
>> >  ----- Original Message -----
>> >  >  Hi,
>> >  >
>> >  >  I'm a newbie to mpi, and i've managed to compile, link and run
>> >  >  my
>> >  >  program under mpich2-1.3.2p1
>> >  >  without any problems. However, when i do the same, but using the
>> >  >  "-mpe=mpilog" option, the following
>> >  >  error is produced:
>> >  >
>> >  >  Writing logfile....
>> >  >  Enabling the Default clock synchronization...
>> >  >  clog_commset.c:CLOG_CommSet_sync_IDs() -
>> >  >  The parent CLOG_CommSet_t does not contain 0-th GID
>> >  >  1188451318-1305386600.768476-degas.como.group in the
>> >  >  child_table[]/nclog_commset.c:CLOG_CommSet_merge() -
>> >  >  CLOG_CommSet_sync_IDs() fails!
>> >  >  Backtrace of the callstack at rank 2:
>> >  >  At [0]:
>> >  >  LINUXlib/mpsimpleprog_vienna(CLOG_Util_abort+0x73)[0x6e2377]
>> >  >  At [1]:
>> >  >  LINUXlib/mpsimpleprog_vienna(CLOG_CommSet_merge+0x2d2)[0x6dec1f]
>> >  >  At [2]:
>> >  >  LINUXlib/mpsimpleprog_vienna(CLOG_Merger_sort+0x1d)[0x6e9c44]
>> >  >  At [3]:
>> >  >  LINUXlib/mpsimpleprog_vienna(CLOG_Converge_sort+0x20)[0x6e631e]
>> >  >  At [4]:
>> >  >  LINUXlib/mpsimpleprog_vienna(MPE_Finish_log+0x114)[0x6e2144]
>> >  >  At [5]:
>> >  >  LINUXlib/mpsimpleprog_vienna(MPI_Finalize+0x58c)[0x6b228a]
>> >  >  At [6]:
>> >  >
>> >  LINUXlib/mpsimpleprog_vienna(_ZN5boost3mpi11environmentD1Ev+0x63)[0x69dc53]
>> >  >  At [7]: LINUXlib/mpsimpleprog_vienna(main+0x56e)[0x5909d6]
>> >  >  At [8]: /lib64/libc.so.6(__libc_start_main+0xe6)[0x3e4c21e576]
>> >  >  At [9]: LINUXlib/mpsimpleprog_vienna[0x58fca9]
>> >  >  application called MPI_Abort(MPI_COMM_WORLD, 1) - process 2
>> >  >  [mpiexec at degas.como.group] ONE OF THE PROCESSES TERMINATED
>> >  >  BADLY:
>> >  >  CLEANING UP
>> >  >  APPLICATION TERMINATED WITH THE EXIT STRING: Terminated (signal
>> >  >  15)
>> >  >
>> >  >  Please help!
>> >  >
>> >  >  Thanks,
>> >  >
>> >  >  Peter
>> >  >
>> >  >  _______________________________________________
>> >  >  mpich-discuss mailing list
>> >  >  mpich-discuss at mcs.anl.gov
>> >  >  https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>
>>
>> --
>> --------------------------------------------------------------------------
>> Peter Man University of Cambridge,
>> Department of Chemical Engineering
>> and Biotechnology,
>> Tel. +44 (0)1223 334786 Pembroke Street,
>> Fax. +44 (0)1223 334796 Cambridge CB2 3RA, UK
>>
>> email: mailto:plwm2 at cam.ac.uk
>> mailto:plwm2 at cantab.net
>> --------------------------------------------------------------------------
>>
>>
>> _______________________________________________
>> mpich-discuss mailing list
>> 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
>



-- 
Mandar Gurav
http://www.mandargurav.org