You did not say what problem you are having.
Before you go any further, you should probably read the Heterogeneity
section of the LAM FAQ -- it contains a lot of useful information on this
topic.
Keep in mind that running in heterogeneous environments (such as the mixed
cluster than you mentioned) can be complicated, and may not produce the
speedups that you want -- you may be limited to the slowest machine(s),
depending on your software, etc.
On Sun, 28 Dec 2003, Dev Anand wrote:
> I ahve a cluster with the following configuration:
> 0-12 Redhat 9 on AMD Athlon 1.2 Ghz
> 13-15 Intel Xeon 2.8 Ghz
> 16 Pentium II 330 Mhz
>
> I can run the following
> mpirun -v n13-15 -s n16 synbench 8192 10 512
> mpirun -v -np 13 -s n16 synbench 8192 10 512
>
> but have problem running this
> mpirun -v -np 16 -s n16 synbench 8192 10 512
>
> synbench is a simple simulation of a nearest neighbour program.
>
> details:
> ___________________________________________________
> mpirun -v n13-15 -s n16 synbench 8192 10 512
> 15510 synbench running on n13
> 20786 synbench running on n14
> 32429 synbench running on n15
> ProbSize: 8192,chunksize 4096, Processes: 2, iter: 10, compfactor: 512
> id: 1, previous: 0, next: 2
> id: 2, previous: 1, next: 0
> 8192 4096 2 10 512 0.076542
>
> mpirun -v -np 13 -s n16 synbench 8192 10 512
> 28552 synbench running on n0
> 820 synbench running on n1
> 26971 synbench running on n2
> 32061 synbench running on n3
> 18919 synbench running on n4
> 25656 synbench running on n5
> 9540 synbench running on n6
> 25775 synbench running on n7
> 19614 synbench running on n8
> 18981 synbench running on n9
> 19431 synbench running on n10
> 19494 synbench running on n11
> 28731 synbench running on n12
> ProbSize: 8192,chunksize 682, Processes: 12, iter: 10, compfactor: 512
> id: 4, previous: 3, next: 5
> id: 1, previous: 0, next: 2
> id: 8, previous: 7, next: 9
> id: 2, previous: 1, next: 3
> id: 3, previous: 2, next: 4
> id: 6, previous: 5, next: 7
> id: 10, previous: 9, next: 11
> id: 7, previous: 6, next: 8
> id: 9, previous: 8, next: 10
> id: 5, previous: 4, next: 6
> id: 11, previous: 10, next: 12
> id: 12, previous: 11, next: 0
> 8192 682 12 10 512 1.747621
>
> mpirun -v -np 16 -s n16 synbench 8192 10 512
> 28553 synbench running on n0
> 821 synbench running on n1
> 26972 synbench running on n2
> 32062 synbench running on n3
> 18920 synbench running on n4
> 25657 synbench running on n5
> 9541 synbench running on n6
> 25776 synbench running on n7
> 19615 synbench running on n8
> 18982 synbench running on n9
> 19432 synbench running on n10
> 19495 synbench running on n11
> 28734 synbench running on n12
> 15532 synbench running on n13
> 20805 synbench running on n14
> 32448 synbench running on n15
>
>
>
>
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
