VMD-L Mailing List
From: John Stone (johns_at_ks.uiuc.edu)
Date: Mon Aug 28 2006 - 01:21:49 CDT
- Next message: alexandra.marques_at_fc.up.pt: "water numbering"
- Previous message: L. Michel Espinoza-Fonseca: "Re: autoionize behaviour on big systems"
- In reply to: L. Michel Espinoza-Fonseca: "Re: autoionize behaviour on big systems"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Michel,
On Mon, Aug 28, 2006 at 01:12:18AM -0500, L. Michel Espinoza-Fonseca wrote:
> Hi John,
> You're right, I was just wondering how well autoionize will perform,
> but I see that it is VERY slow, just to share that with you guys :).
For small systems autoionize runs ok, but all kinds of things become
challenging when you're in the 8 million atom realm. :)
We wrote the new program because (even besides the autoionize plugin, and
some other C-based ion placement code we had already) we didn't
know of any existing ion placement tool that would handle multi-million atom
systems in less than a week of runtime on the machines available to us.
> I'll definitely prepare my system with the new ionize tool. I'll make
> you know what I got.
Even with the parallel ionize tool it's going to take some runtime for the
8 million atom structure... I'll be curious to hear how it goes for you.
> Thanks!
> Michel
>
> 2006/8/28, John Stone <johns_at_ks.uiuc.edu>:
> >
> >Michel,
> > Yes, now you see why I suggested using the new tool... ;-)
> >The existing autoionize plugin is just a Tcl script, and so it doesn't
> >have the ability to use more than one processor (even though VMD itself
> >can). The new ionize program we have written is compiled C code,
> >and thus it is able to run on multiple processors and runs much faster
> >than any scripting language is presently able to. Placing ions on
> >multi-million atom systems on a single processor can take _several days_
> >of CPU time, even with reasonably written C code. The new ionize tool
> >we've written uses both vectorization (use the Intel C/C++ compiler)
> >and multithreading in order to run much much faster. The new ionize
> >program
> >can use all 48 processors of your Altix to good effect, the script-based
> >autoionize plugin cannot.
> >
> >Thanks,
> > John
> >
> >On Mon, Aug 28, 2006 at 12:44:24AM -0500, L. Michel Espinoza-Fonseca wrote:
> >> Hi all,
> >>
> >> I've been testing the autoionize plugin using VMD 1.8.5 and a 8M atoms
> >> system. Even though I haven't tried the ionize program yet, I wanted
> >> to see how autoionize behaves with such a big system. I ran autoionize
> >> on a 48-processor Altix machine with 96 GB of RAM, and autoionize has
> >> been working for more than a day and still doesn't finish! I'm not
> >> complaining at all, I'm just sharing what I'm getting. It might be
> >> useful to mention that VMD is using only 1 processor out of the 48
> >> available, and the memory requirements averages 6 GB.
> >>
> >> That's all for now, I'll keep testing it and I'll make you know about
> >> the results.
> >>
> >> Enjoy!
> >>
> >> Michel
> >
> >--
> >NIH Resource for Macromolecular Modeling and Bioinformatics
> >Beckman Institute for Advanced Science and Technology
> >University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
> >Email: johns_at_ks.uiuc.edu Phone: 217-244-3349
> > WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
> >
-- NIH Resource for Macromolecular Modeling and Bioinformatics Beckman Institute for Advanced Science and Technology University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801 Email: johns_at_ks.uiuc.edu Phone: 217-244-3349 WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
- Next message: alexandra.marques_at_fc.up.pt: "water numbering"
- Previous message: L. Michel Espinoza-Fonseca: "Re: autoionize behaviour on big systems"
- In reply to: L. Michel Espinoza-Fonseca: "Re: autoionize behaviour on big systems"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]