Fw: [MITgcm-support] Scalability

Stefano Querin squerin at ogs.trieste.it
Tue Aug 9 09:49:12 EDT 2005


Hi,
I forward to all the list the message I sent to Chris some days ago. I also 
attach the results of a recent run (a different simulation) with 8 procs: 
system time is huge and almost all the gap between user+system and wall 
clock time is created by DYNAMICS, SOLVE_FOR_PRESSURE and BLOCKING_EXCHANGES 
routines!
Is the network of our cluster too slow? Can it be a matter of code 
optimization or compiler options?
Thanks a lot!

Stefano


----- Original Message ----- 
From: "Stefano Querin" <squerin at ogs.trieste.it>
To: <cnh at mit.edu>
Sent: Wednesday, August 03, 2005 5:02 PM
Subject: Re: [MITgcm-support] Scalability


> Hi Chris,
> I forgot the details of the connection, you're right (sorry...)!
> The cluster has an ethernet copper wire network (1Gb). The controller is a 
> Broadcom BCM5703 10/100/1000BASE-T Gigabit Controller. The switch is hp 
> procurve 2824 
> (http://www.hp.com/rnd/products/switches/2800_series/overview.htm).
> If you want to access to the system there is NO problem at all, just let 
> me ask to the system administrator how to do that. I think that from 
> outside the LAN of our institute, people should connect to a local server 
> and then ssh to the cluster. I will check it as soon as possible and (if 
> you want) make an account for you.
> Thanks a lot!
>
> Stefano
>
>
> ----- Original Message ----- 
> From: "chris hill" <cnh at mit.edu>
> To: <mitgcm-support at mitgcm.org>
> Sent: Wednesday, August 03, 2005 3:57 PM
> Subject: Re: [MITgcm-support] Scalability
>
>
>> Stefano,
>>
>>  Sorry to hear that it wasn;t a simple I/O thing.
>>  What sort of network do you have connecting the nodes in the cluster?
>>  Not sure what the problem is - it could be the network connecting the
>> nodes or it could be a system configuration thing.
>>  If you can give us access to the system we can probably figure out
>> what the cause is and whether it is easy to fix.
>>
>> Thanks,
>>
>> Chris
> 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: STDOUT.txt
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20050809/b9dc5113/attachment.txt>


More information about the MITgcm-support mailing list