[mpich-discuss] mpe clog2 file size questions

[Anthony Chan]

The file size changes mainly because it contains different number
of clog2 records.  So it's either one run is longer than the other
or the number of processes are different.  You will know when you
view them with jumpshot.

----- Original Message -----
> Thx for reply.
> yeah, jumpshot will convert clog to a smaller slog file. but large
> file has
> too much overhead to app performance.
> Can you anwer the question: "Does it normal of the logfile size? why
> does
> the file size vary so much?"
> 
> On Wed, May 2, 2012 at 11:35 AM, Anthony Chan <chan at mcs.anl.gov>
> wrote:
> 
> >
> > Why don't you convert them into slog2 files and view them with
> > jumpshot
> > to see how they are different.
> >
> > A.Chan
> >
> > ----- Original Message -----
> > > Hi all:
> > > I try to use mpe in HPL benchmark with compile option -mpe=mpilog.
> > > When i run with 4 processes, it generate a clog2 file about 200M.
> > > And
> > > the
> > > file size varies when i run it twice. Does it normal of the
> > > logfile
> > > size? I
> > > mean it's too big. I ever used IPM of which the log file is fairly
> > > small.
> > > And why does the file size vary so much?
> > >
> > > Below is the log file size:
> > >
> > > -rw-rw-r-- 1 jujj jujj 197068180 2012-05-02 09:15 old.xhpl.clog2
> > > //
> > > first
> > > run
> > > -rw-rw-r-- 1 jujj jujj 16254356 2012-05-02 09:16 xhpl.clog2 //
> > > second
> > > run，only one-tenth of the first one
> > >
> > > _______________________________________________
> > > mpich-discuss mailing list mpich-discuss at mcs.anl.gov
> > > To manage subscription options or unsubscribe:
> > > https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
> > _______________________________________________
> > mpich-discuss mailing list mpich-discuss at mcs.anl.gov
> > To manage subscription options or unsubscribe:
> > https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
> >