[Wrf-users] Wrf-users Digest, Vol 136, Issue 9

[sam.hawkins at vattenfall.com]

Hi Tabish Ansari,

How well the performance of WRF scales depends on a number of factors, but it is possible to get good scaleability well beyond 120 processors.  Mainly it depends on:

1. Domain size
The domains get decomposed into tiles and patches for separate processors to work on. If these become too small, then each processors does not have enough to do, and spends more time waiting for messages from neighbouring domains.  

2. The hardware
Particularly the communication infrastructure e.g. infiniband. If there is a high latency, then this will prevent good scaling.  Disk writes can also be bottleneck as well, which will prevent scaling. 

3. Type of parallelism
WRF supports shared memory (SMP) and distributed memory parallelism (MPI) , and combinations of both. These are selected at compile-time.  I've found the simplest and (usually) best performance is distributed memory parallelism, unless you *really* want to squeeze the last ounce of performance out of WRF.

There are some good test cases here http://www2.mmm.ucar.edu/wrf/WG2/benchv3/  which use a large domain, I would suggest testing those on and seeing how your performance compares. Use a small number of processors first, and gradually scale up, plotting the results, that way you can see where the performance starts to tail off.


Sam Hawkins

-----Original Message-----
From: wrf-users-bounces at ucar.edu [mailto:wrf-users-bounces at ucar.edu] On Behalf Of wrf-users-request at ucar.edu
Sent: Wednesday, December 16, 2015 7:00 PM
To: wrf-users at ucar.edu
Subject: Wrf-users Digest, Vol 136, Issue 9

Send Wrf-users mailing list submissions to
	wrf-users at ucar.edu

To subscribe or unsubscribe via the World Wide Web, visit
	http://mailman.ucar.edu/mailman/listinfo/wrf-users
or, via email, send a message with subject or body 'help' to
	wrf-users-request at ucar.edu

You can reach the person managing the list at
	wrf-users-owner at ucar.edu

When replying, please edit your Subject line so it is more specific than "Re: Contents of Wrf-users digest..."


Today's Topics:

   1. Scaling ability of WRF and WRF/Chem (Tabish Ansari)


----------------------------------------------------------------------

Message: 1
Date: Tue, 15 Dec 2015 01:18:31 +0000
From: Tabish Ansari <tabishumaransari at gmail.com>
Subject: [Wrf-users] Scaling ability of WRF and WRF/Chem
To: "wrfchemhelp.gsd at noaa.gov" <wrfchemhelp.gsd at noaa.gov>,	wrfhelp
	<wrfhelp at ucar.edu>, "wrf-users at ucar.edu" <wrf-users at ucar.edu>
Message-ID:
	<CALLVTyvBDfn=eMMa2-J+j9Ckv6Y4qUe5KEPH-MUOBNnuh=i3-w at mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Hi

I am trying to submit a WRF/Chem job on 120 processors (15 nodes x 8 cores) but it wouldn't scale fully.

What's the typical scalability of WRF?

Thanks

Tabish

Tabish U Ansari
PhD student, Lancaster Environment Center Lancaster Univeristy Bailrigg, Lancaster,
LA1 4YW, United Kingdom
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/wrf-users/attachments/20151215/44bb26f5/attachment-0001.html 

------------------------------

_______________________________________________
Wrf-users mailing list
Wrf-users at ucar.edu
http://mailman.ucar.edu/mailman/listinfo/wrf-users


End of Wrf-users Digest, Vol 136, Issue 9
*****************************************

#======================================================================#
We have recently changed the registered offices of a number of our companies. The following are now registered at 1 Tudor Street, London, EC4Y 0AH:
Vattenfall Wind Power Ltd, Border Wind Ltd, Border Wind Farms Ltd, BW Ops Ltd, Clashindarroch Wind Farm Ltd,  Eclipse Energy UK Ltd, 
Eclipse Energy Company Ltd, Kentish Flats Ltd, Ormonde Energy Ltd, Ormonde Energy Holdings Ltd, Ormonde Project Company Ltd, Thanet Offshore