[gmx-users] Slow opterons : low Mega Flops
David van der Spoel
spoel at xray.bmc.uu.se
Tue Sep 13 08:30:15 CEST 2005
On Mon, 2005-09-12 at 14:54 -0700, Gia Maisuradze wrote:
> Thanks Erik, actually it turns out I was using gcc not ifc, now gromacs is
> running with the expected amount of speed.
Compare your numbers to the benchmarks. For me gcc is faster than icc
8.0. Fortran is not necessary (slower) on intel/amd.
> Thanks again.
> ----- Original Message -----
> From: "Erik Lindahl" <lindahl at sbc.su.se>
> To: "Discussion list for GROMACS users" <gmx-users at gromacs.org>
> Sent: Monday, September 12, 2005 1:32 PM
> Subject: Re: [gmx-users] Slow opterons : low Mega Flops
> > Hi,
> > We need details, details, details :-)
> > 1. Skip MPI when it comes to basic performance testing.
> > 2. Compile FFTW 2.1.5 and Gromacs (preferrably the new release candidate
> > at ftp://ftp.gromacs.org/pub/beta/ ) with default options.
> > 3. Download the standard benchmarks from www.gromacs.org
> > 4. If you get results that seem to differ significantly from the ones
> > quoted on the benchmark page, report with your CPU type and frequency
> > here, and the exact results you get.
> > Remember to say whether you are running the operating system in 32 or 64
> > bit mode, and check the logfile whether it is using assembly loops.
> > Check that you are not running other heavy jobs in the node when
> > testing - Gromacs defaults to nice 19 (only use idle cycles), which
> > means it's going to yield to almost everything else. At the end of the
> > run you'll get a report both of the actual time, and the time used by
> > Gromacs.
> > Cheers,
> > Erik
> > On Sep 12, 2005, at 9:53 PM, Gia Maisuradze wrote:
> >> Hi,
> >> I'm getting really slow performance on my opteron system, performance
> >> that is roughly half that of my old athlon mp system. I've tried
> >> recompiling the program with various options, mpi, no mpi, threading, no
> >> threading, cpu optimized, not optimized, etc. Does anyone work with
> >> opterons under suse 9.3 and have an idea about what the best options for
> >> compiling lam, fftw, and gromacs should be?
> >> Thanks,
> >> Gia Maisuradze
> >> _______________________________________________
> >> gmx-users mailing list
> >> gmx-users at gromacs.org
> >> http://www.gromacs.org/mailman/listinfo/gmx-users
> >> Please don't post (un)subscribe requests to the list. Use the www
> >> interface or send it to gmx-users-request at gromacs.org.
> > _______________________________________________
> > gmx-users mailing list
> > gmx-users at gromacs.org
> > http://www.gromacs.org/mailman/listinfo/gmx-users
> > Please don't post (un)subscribe requests to the list. Use the www
> > interface or send it to gmx-users-request at gromacs.org.
> gmx-users mailing list
> gmx-users at gromacs.org
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596, 75124 Uppsala, Sweden
phone: 46 18 471 4205 fax: 46 18 511 755
spoel at xray.bmc.uu.se spoel at gromacs.org http://xray.bmc.uu.se/~spoel
More information about the gromacs.org_gmx-users