New virginia tech supercomputer to fold?

[nahmus]

here is a few clips from an article I just got

RICHMOND, Va. (AP) - A supercomputer made from 1,100 off-the-shelf Apple Macs at Virginia Tech now ranks third among the world's 500 fastest supercomputers, many of which handle with ease 1 trillion calculations per second.

The Terascale Cluster computer, nicknamed "Big Mac" by some, conducts 9.6 trillion calculations per second, zipping along with 2,200 IBM microprocessors not unlike those found in high-end consumer desktop computers.

Virginia Tech plans to unleash Big Mac on the school's growing research interest in such areas as nanoelectronics, aerodynamics and the molecular modeling of proteins.




ok now I have 2 questions. Does anybody here go to that school and how do we get them folding for team 32!!!!

__

[Ducker]

ok now I have 2 questions. Does anybody here go to that school and how do we get them folding for team 32!!!!

Wow.........that would be sweet if we could get them on our team!!! :cool:

[Loud]

<reality check>

FAH only allows four (count 'em ... 4) instances to run per machine. That would leave 1,096 CPUs idle.

[Cowboy Shane]

4.0 beta client allows 8 :D, but the client doesn't really support clusters from what I understand. I guess you could have all 1,100 nodes folding as individual computers.

[Fast420A]

They are probably using a different program than F@H to use the 1100 CPU cluster.

[OCn00b]

I live about 2 hours away from Tech but don't have time to travel down there. I'm sure there are a few members here that live in the Blacksburg area that could stop by.

[NiTrO bOiE]

Originally posted by Cowboy Shane
4.0 beta client allows 8 :D, but the client doesn't really support clusters from what I understand. I guess you could have all 1,100 nodes folding as individual computers.

How'd you find that out?

[Cowboy Shane]

I found out about the beta and its features in these very forums (http://forum.oc-forums.com/showthread.php?s=&threadid=243545)

[nahmus]

I would love to send an e-mail if I could figure out who to send it to.

[Cowboy Shane]

This is from 1996 (http://scholar.lib.vt.edu/vtpubs/spectrum/sp961205/1b.html), but it lists Peter Athanas and Ray Bittner of Virginia Tech's Department of Electrical and Computing Engineering as having built an earlier model of something like the Big Mac. Perhaps if they didn't take part in this one, one of them knows who did.

Edit: Here is the faculty page of Dr. Athanas (http://www.ece.vt.edu/faculty/athanas.html), but if you are serious about contacting them, you will probably want to contact the department head, Warren Stutzman, or the System Administrator, John Harris. You can get their contact info from this page (http://www.ecpe.vt.edu/people/index.html).

[nahmus]

I just sent an e-mail to the sys admin (thanks cowboy shane)inviting him to come to the forums and if the type of protein modeling that they are doing is the same as folding@home to take a look at our team and perhapse consider joining.

Hmmm I think i better clean up a bit! hahah :D

does F@H run on mac's?

I did not see what OS they are using.

I told them that they could use my folding name as a test :D

[nahmus]

I ACTUALLY GOT A REPLY!!!!! (the good news)

now for the bad...

I talked to the administrator and he said that they are still working out some kinks in the system and don't expect it to be on line 24/7 for a few months. He said that the list of projects that would like to be considered for processing is already quite long but he would happily submit sonething for us. I don;t really think that we have a shot BUT he did say that the university does run a grid computing application on all of the college PC's and that he will get me the address of the administrator of the grid computing services to see if we can get TEAM32 added to the grid computing.

so its not all bad. I'll keep up with any developments.

[Seal]

Originally posted by nahmus
ok now I have 2 questions. Does anybody here go to that school and how do we get them folding for team 32!!!!

The answers simple, borg them.

[BBigJ]

Guys, there are a lot of people working on the folding problem other than the Pande group. It is a very difficult problem and many people have different ideas on how to attack it. The Pande group has chosen a medium that gives them far more computing resources than other research groups have at their disposal. However this doesn't mean they have the "best" way of attacking the problem. It is very likely that someone will come out will an algorithm that is more efficient or accurate than what fah is using. However, this will never be discovered unless that someone has the computer resources to test their methods.

In other words, using this computer for fah would be like stealing from the poor to give to the rich. There is no guarantee that the rich have the best product, so let the little guys try to come up with something better (Microsoft vs. Linux would be a good analogy here).

I'm sure that when a better algorithm comes out, Pande and the gang will quickly arrange a collaboration so that they can use the improved code on fah (this will also be to the advantage of the people that invented the new code, as they will have access to the computer resources of fah.)

SBs

[Thespis377]

To run it on a cluster that size you have to install a copy on each node. Now if we could convince someone to write a multi-threaded F@H core....Oh the possiblities!

[NiTrO bOiE]

Here's a pic of the cluster.

http://www.dslreports.com/r0/download/458905~d492cfb99c6bade1227ed1b079f86ed5/IMG_2354.jpg

[sautegod]

I might have room for that.

yo wait sorry the closets full.

any one with an open warehouse.

[zachj]

I don't know about new Macs (the G5 architecture is different than the G4, and the MHz is higher), but G4s absolutely SU** at folding, and I really mean that. I'm using one right now, and I highly doubt it could even turn in a WU on time.

I'm serious . . .

Z

[TommyHolly]

LOL I think the cluster is saying "Iiiii Looooove youuuuuuu" to me?