[gmx-users] benchmark numbers

[vipin sachdeva]

Woodcrest runs a 64 bit OS and compiles natively to a 64 bit binary. I am 
2X slow, for instance the PPC 970 1600 mhz with 512 Kb L2 Cache (Mine is 
2000 mhz dual-chip  with 512 Kb each with one core disabled) gcc number 
for d.villin is 9443  ps/day == 9.443 ns/day. For me the number printed at 
end of GROMACS run is  more like 5 ns/day. For woodcrest the number is 48 
for same benchmark  while I get close to  23. I am wondering why I should 
be this off. Do you  have the binary with you to reproduce these numbers ? 
I can download the binary and run it on my machine, to see if it's a 
problem with the compile options or the machine performance (though I dont 
understand why each one of these should be so off).  thanks for looking 
into this.

Regards
vipin



On Mon, 9 Apr 2007, David van der Spoel wrote:

> vipin sachdeva wrote:
>> 
>> Yes, I am really off (more than 2X for both Woodcrest and PPC970).
>> I was  earlier doing the configure flags
>> 
>> CC=gcc CFLAGS=-O3 LDFLAGS=-O3 --enable-ppc-altivec (--enable-x86-64-sse) 
>> --enable-cpu-optimizations --enable-float --disable-threads (single node 
>> results) only.
>> 
>> and I found I was 2X off. I am just downloading gmxbench, doing grompp 
>> inside d.villin and then just mdrun. Any more options needed for more 
>> optimal performance ? Do flags for software-sqrt or static binaries only 
>> also help a lot ? Both machines are for personal usage only.
>> 
>> 
>
> Does the woodcrest box run a 64 bit OS?
> For PPC970 the IBM compilers could help.
> What does it mean to be 2X off? Too fast or too slow?
>
> -- 
> David.
> ________________________________________________________________________
> 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://folding.bmc.uu.se
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the www interface 
> or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>