[Pw_forum] Predicting the cpu time usage for bigger system withour doing the simulation

Michael Sullivan Michael at ihpc.a-star.edu.sg
Wed Jul 21 07:23:15 CEST 2010


Hi,

I chose N=5 because the original post referred to 5 supercells. If your system is twice as big it will be 2**3, or 8 times bigger.

For the second part of your message, I don't have a good way to know in advance how it's going to perform. Usually I run some tests and build up experience.

If I understand you clearly, I would guess you're running out of memory when you run on a smaller number of nodes. If you're using pools for k-point parallelization, the memory doesn't scale.

Mike
michael at ihpc.a-star.edu.sg<mailto:michael at ihpc.a-star.edu.sg>
http://www.sullivan.sg/

From: pw_forum-bounces at pwscf.org [mailto:pw_forum-bounces at pwscf.org] On Behalf Of psavita at crlindia.com
Sent: Wednesday, July 21, 2010 12:51 PM
To: pw_forum at pwscf.org
Subject: Re: [Pw_forum] Pw_forum Digest, Vol 37, Issue 46 : Re: Predicting the cpu time usage for bigger system withour doing the simulation

Hello Mike,

"DFT usually scales as N^3 so I'd guess that it will take 5**3=125 times longer than the small system. You would probably gain somewhere due to fewer k-Points, etc. but that's my rough estimate."

Regarding your above comment, may I know how you choose N=5, is it just an example?
I have similar related question.
Is it possible to know the most appropriate compute resources for running a simulation without actually carrying out the same? My experience on a cluster is that by specifying a smaller number of nodes than the demand often results into node failure. Can this be avoided?

Thanks in advance for any clarification.

Regards,
Savita Pundlik
Applications Group : Computational Materials Simulations
Computational Research Laboratories Ltd.,
Taco House, Damle Path, Off Law College Road
Pune - 411004, India.


________________________________
This email is confidential and may be privileged. If you are not the intended recipient, please delete it and notify us immediately. Please do not copy or use it for any purpose, or disclose its contents to any other person. Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20100721/150986f4/attachment.html>


More information about the users mailing list