[gmx-users] mdrun load sharing between dual processor machine

[David van der Spoel]

On Thu, 2 May 2002, Oliver Beckstein wrote:

>On Thu, 2 May 2002, Lynne E. Bilston wrote:
>
>> Also, does ssh slows down the interconnect compared to rsh?
>
>According to the LAM documentation ssh/rsh is only used to start the
>processes on the other nodes; the actual communication is done through lam
>and is determined by the -c2c or -lamd options to mpirun (-c2c is supposed
>to be faster). The underlying transport mechanism that is determined at
>compile time (--with-rpi=tcp|sysv|usysv) might also influence your
>performance. usysv seems to be recommended if it works on your system but
>I haven't done any proper tests.
>
>[Has anyone looked at the influence of these variables on performance? 
>Any recommendations--esp. for clusters of Athlon duals?]
Yes:
http://zorn.bmc.uu.se/~spoel/scaling

that was with gromacs 2 before dual athlons were invented...

>Oli
>
>

-- 
Groeten, David.
________________________________________________________________________
Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
Husargatan 3, Box 576,  	75123 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++