The ABINIT Users Mailing List ( CLOSED )

[Marcel Mohr <marcel@physik.tu-berlin.de>]

without parallelization over k-points, it is getting worser: 
timer: Routine,Calls,Time,% = IterSCF        1    1988.396  97.37
elaps: Routine,Calls,Wall,% = IterSCF        1     545.233  97.17

On Tue, 28 Aug 2007, Marcel Mohr wrote:

> Another option:
> We use dual xeons on a dual motherboard.
> Here is an example output from the LCAO DFT code SIESTA (I have no parallel 
> abinit output.)
> I used parallelization over 8  k-points.
>
> timer: Routine,Calls,Time,% = IterSCF        1    2377.865  97.82
> elaps: Routine,Calls,Wall,% = IterSCF        1     605.648  97.42
>
> In the 1. line is the cpu time, below is the elapsed time.
> Here the time almost scales linearly with number of processors. (3.92 instead 
> of 4)
>
> Cheers Marcel
>
>
>
>
> On Tue, 28 Aug 2007, Xiulin Ruan wrote:
>
> > Thanks for the information. We tested quad-cores and we do find the
> > performance is no better than dual-cores. So we will go with dual-core
> > processors. One more question. It seems AMD processors have better
> > performance than Intel for abinit, right?
> >
> > Thanks,
> >
> > Xiulin
> >
> >
> > > -----Original Message-----
> > > From: Nicola Marzari [mailto:marzari@mit.edu]
> > > Sent: Saturday, August 18, 2007 2:50 PM
> > > To: forum@abinit.org
> > > Subject: Re: [abinit-forum] computer cluster for parallel computing
> > >
> > >
> > >
> > > Also, make sure you really need quad-cores. I've tested other plane-wave
> > > electronic structure codes, and never found a case when the quad cores
> > > were justified, for the marginal increase of performance compared to two
> > > cores - you usually saturate the memory bus very quickly, even with one
> > > core.
> > >
> > > This is even more relevant when you factor in the true costs of air
> > > conditioning and electricity.
> > >
> > >                                 nicola
> > >
> > >
> > > Xiulin Ruan wrote:
> > > > Thanks a lot for the information. Also, what is the proper amount of
> > > memory
> > > > per core? Is 4GB good enough?
> > > >
> > > > Thanks,
> > > >
> > > > Xiulin
> > > >
> > > > > -----Original Message-----
> > > > > From: Anglade Pierre-Matthieu [mailto:anglade@gmail.com]
> > > > > Sent: Friday, August 17, 2007 3:41 PM
> > > > > To: forum@abinit.org
> > > > > Subject: Re: [abinit-forum] computer cluster for parallel computing
> > > > >
> > > > > Hi,
> > > > >
> > > > > Two partial answers for your question:
> > > > > - I have never done any benchmarking but infiniband works very well
> > > > > for Abinit on "lemaitre" the super computer of the CISM
> > > > > - You will find a detailed performance analysis by Torsten Hoefler
> > > > > with two different kinds of interconnection here :
> > > > > www.unixer.de/publications/img/hoefler-sfb05.pdf
> > > > > This might be usefull for your decision.
> > > > >
> > > > > Regards
> > > > >
> > > > > PMA
> > > > >
> > > > > On 8/17/07, Xiulin Ruan <ruan@purdue.edu> wrote:
> > > > > > Dear all,
> > > > > >
> > > > > > We are building a computer cluster to run abinit in the parallel mode.
> > > > > We
> > > > > > are gonna have two or four nodes, and each node will be
> > dual-processor,
> > > > > > quad-core. We wonder if the high speed InfiniBand connection is worthy
> > > > > to
> > > > > > have. Our current jobs include calculating band structure and optical
> > > > > > properties of carbon nanotubes and TiO2. Any suggestions?
> > > > > >
> > > > > > Many thanks,
> > > > > >
> > > > > > Xiulin
> > >
> > >
> > > --
> > > ---------------------------------------------------------------------
> > > Prof Nicola Marzari   Department of Materials Science and Engineering
> > > 13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
> > > tel 617.4522758 fax 2586534 marzari@mit.edu http://quasiamore.mit.edu