[GE users] Yet anaother MPICH tight-integration problem

John Hearns john.hearns at clustervision.com
Tue Sep 7 06:58:20 BST 2004


    [ The following text is in the "iso-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

On Tue, 2004-09-07 at 05:07, David S. wrote:
> I'm installed SGE V60u1 and MPICH 1.2.6 in x86 Linux 2.4.  I've got 
> a cluster of 39 dual processor boxes.  I'm using the "ch_p4" MPICH 
> device over ethernet,
David,
I am deploying a similar setup this week, with mpich 1.2.6
When I get things set up I will run similar tests, and report back.
I'd be concerned also if there was some sort of limit.

I just tried a test job with SGE 6.0u1, x86 Linux 2.4 
and MPICH 1.2.5.2-1
This cluster has 47 slave nodes, each dual processor.
I'm only getting 47 slots when I request a range of 2-90

I won't cry wolf at this stage - as the list knows, I've been
rearranging slot allocations on the queues and hosts,
and yesterday I changed the allocation rule on the PE to
$round_robin in response to a request.

I'll try a few more test jobs.


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list