Thanks a lot Dr. Stefano for your reply.<br><br><div class="gmail_quote">On Fri, Dec 24, 2010 at 3:50 PM, Stefano de Gironcoli <span dir="ltr"><<a href="mailto:degironc@sissa.it">degironc@sissa.it</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">It strongly depends on the particular calculation and on the<br>
parallelization strategy you choose. Linear scaling would be the ideal<br>
scaling which you never get.<br>
The things to keep in mind are load balancing and communication<br>
overhead. This last is affected by your communication network latency<br>
and bandwidth ... As a general rule the bigger the calculation the<br>
easier is to scale.<br>
It may be useful to have a look at the final timing summary of a<br>
calculation to see the fraction of time spent in communication (FFT<br>
scatter/gather and reduce operations) compared with the total wall clock<br>
time for your system and how this changes for different settings.<br>
Another factor to keep in mind is RAM memory as with certain<br>
parallelization strategies you can trade some speed with increased memory.<br>
<br>
stefano<br>
<div class="im"><br>
mohnish pandey wrote:<br>
> Dear QE users,<br>
> I am trying to see how the runtime scales with<br>
> number of processors. I did the same calculation using one node with eight<br>
> processors on one cluster and 8 nodes with total 56 processors on other<br>
> cluster but the time does not seem to scale linearly with the number of<br>
> processors. Can anybody give me an idea how does the time scale with number<br>
> of processors.<br>
> Thanks a lot in advance.<br>
><br>
><br>
</div>> ------------------------------------------------------------------------<br>
><br>
> _______________________________________________<br>
> Pw_forum mailing list<br>
> <a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
> <a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>
><br>
<br>
_______________________________________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
<a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>
</blockquote></div><br><br clear="all"><br>-- <br>Regards,<br>MOHNISH,<br>-----------------------------------------------------------------<br>Mohnish Pandey<br>Y6927262,5th Year dual degree student,<br>Department of Chemical Engineering,<br>
IIT KANPUR, UP, INDIA<br>-----------------------------------------------------------------<br>