[mpich-discuss] MPI_WTime

[Rajeev Thakur]

It uses different options on different platforms. See the file
src/mpi/timer/mpidtime.c in MPICH2.
https://svn.mcs.anl.gov/repos/mpi/mpich2/trunk/src/mpi/timer/mpidtime.c

Rajeev

> -----Original Message-----
> From: owner-mpich-discuss at mcs.anl.gov 
> [mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Daniel Becker
> Sent: Saturday, April 19, 2008 7:38 AM
> To: mpich-discuss at mcs.anl.gov
> Subject: [mpich-discuss] MPI_WTime
> 
> Dear MPICH team,
> 
> I am working at Research Center Juelich (Germany) and I am 
> currently evaluating different methods to query clock values 
> on cluster systems. 
> Of course, MPI_Wtime() is one of the options a user has. I am 
> interested how MPICH implements MPI_Wtime() on such cluster systems.
> 
> I guess that MPICH implements MPI_Wtime() in various ways 
> accordingly to the detected system architecture. However, I 
> assume that you have some kind of default mechanism within 
> MPICH to calculate a return value of
> MPI_Wtime() on unknown systems. Moreover, I assume that this 
> default mechanism relies on gettimeofday(). Are my 
> assumptions correct or how is it done? Perhaps you also use NTP?
> 
> I am unsure whether this is the correct mailing list, but it 
> would be great if you could help me. Thanks a lot for your 
> help in advance :-).
> 
> Best regards,
> Daniel
> 
> --
> Dipl.-Ing. Daniel Becker
> Juelich Supercomputing Centre (JSC)
> Forschungszentrum Jülich GmbH
> 
> Phone:  +49-2461-61-3650
> Fax:    +49-2461-61-6656
> E-mail: d.becker at fz-juelich.de
> 
>