[gmx-developers] Benchmark Results Clarification

[David]

On Wed, 2003-08-06 at 23:27, Josh Hursey wrote:
> I am working with a research group attempting to replicate the
> (normalized versions) of you results posed on the gromacs site. We are
> falling short of the ps/day that you have posted.
> 
> To ensure that we are running a close setup to the environment in which
> your tests were ran, Could you provide us with some additional information
> that I could not find on the website.
> 
> We are running:
>   Cluster 1:
>     Dual Pentium III (Katmai) 497 MHz 512 KB L2 Cache
>     512 MB S-DRAM
>     100 MB Switched environment
>     Linux version 2.4.7-10smp
>     gcc version 2.96
>     Red Hat Linux 7.1 2.96-98
>   Cluster 2:
>     Dual PowerPC G4 7455, altivec supported 999MHz 256K L2 cache
>     2 MB at 4 GBps
>     1 GB DDR SDRAM
>     1 GB switched Environment
>     Linux version 2.4.20-8d
>     gcc version 3.2.2
>     Yellow Dog Linux 3.0 3.2.2-2a
> 
> 1) What FFTW configuration options were used on the install?
Nothing special, except the i386 tweaks and single precision

> 
> 2) What Gromacs compiler options were used on the install?
> 
all default as generated by the configure script


> 3) What hardware/networking environment were the benchmarks ran under?
> 
Different ones, but there are results for 100 Mb/s switched system too
We used LAM 6.x for MPI where a few extra flags were used for compiling
-tcp-short-message=500000 -shm-short-message=500000 -usysv
(check for exact syntax).

> 4) Are the MPI tests ran so that a MPI run with 2 processes are run on one
> dual processor node or on 2 different nodes?
>    (e.g.
>      2 processes = 1 node with 2 processors
Yes

>      or
>      2 processes = 2 nodes with 2 processors
>    )
> 
> Thank you for your help!
> 
Furthermore check in the output from your jobs whether the SSE routines
are used on your Pentium 3.

> Peace,
> 
> Josh Hursey
> Earlham College
> 
> _______________________________________________
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-developers-request at gromacs.org.
-- 
Groeten, David.
________________________________________________________________________
Dr. David van der Spoel, 	Dept. of Cell and Molecular Biology
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
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++