Hello,
Works great for me, using lam/mpi 7.0 compiled in 64 bit adressing on ibm
RS6000 using the native xlf/xlc compilers and compiled in 32 bit (of
course) on Linux using intel ifc/icc. This example mixes (big and little
endians, lam takes care automatically of it. Also under lam 7.0, this data
conversion is automatically disabled if your platform is homogeneous.
Nothing special to start heterogeneous runs over a mix of rs6000 and minux
uniprocs and smps, just use the proper application schema to specify the
location of executables in addition to the usual host schema. Great if you
need for instance to combine a large memory for some poorly parallelized
steps and large number crunching the rest of time.
Just be careful to implement some kind of load balancing if your cpus are
of different speed...
Best.
Pierre V.
On 28 Jul 2003, Carlos Lopez Nataren wrote:
> Hello, I was reading the documentation on LAM and it says that it works
> very well on heterogenous clusters, but it does not explain why, does
> anybody knows if LAM has any particular support for heterogenous
> systems???
>
> Any help will be greatly appreciated.
>
> Carlos Lopez
>
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
_/_/_/_/ _/ _/ Dr. Pierre VALIRON
_/ _/ _/ _/ Laboratoire d'Astrophysique (UMR 5571 CNRS)
_/ _/ _/ _/ Observatoire de Grenoble / U. Joseph Fourier
_/_/_/_/ _/ _/ BP 53 F-38041 Grenoble Cedex 9 (France)
_/ _/ _/
_/ _/ _/ http://www-laog.obs.ujf-grenoble.fr
_/ _/ _/ mailto:Pierre.Valiron_at_[hidden]
_/ _/_/ Phone / Fax: +33 (0)4 76.51.47.87 / (0)4 76.44.88.21
|
