Video of the Day: 23 GeForce GTX 295 Video Cards Installed in a Single Server Rig

[newtekie1]

damn! ... I am still confused about how you link up all those pcs, and how do you get them all onto one monitor...?

They aren't really linked, they don't need to be really. They are all on the same network, that is about it.

As for connecting them all the to the same monitor, I'm sure the rack has a KVM, which allows multiple machines to be connected to the same Keyboard, Monitor, and Mouse. Although, once set-up the machines should be able to be run headless, with remote access via LogMeIn or Remote Desktop.

 

[h3llb3nd4]

ahh now I understand...

 

[mdm-adph]

I was expecting a custom, 23 PCI Express slot motherboard board loaded with GPUs (like a Tesla system). Needless to say, I'm disappointed. Anyway, that's quite a blatant waste of money and electricity, in my opinion.

Please go to the guy's site and read his story before you call it a waste.

 

[FordGT90Concept]

There's better ways to hunt down a cure to Huntington's Disease than protein folding simulations through Stanford University. $30,000 towards the development of nanomachines, for example, would be far more beneficial and practical use of the money.

 

[newtekie1]

There's better ways to hunt down a cure to Huntington's Disease than protein folding simulations through Stanford University. $30,000 towards the development of nanomachines, for example, would be far more beneficial and practical use of the money.

Actually, $30,000 would likely not provide any useful funding towards real research. Most of these programs have yearly budgets in the millions, $30,000 would hardly make a difference. For relatively small amounts of money like $30,000, a Folding farm is actually a pretty good way to put it to use.

 

[FordGT90Concept]

Almost all research, aside from pharmaceuticals, is funded by grants and public donations. $30k is enough to hirer a research for 1/3 to 1/4 of a year. 90 to 120 days which amounts to about 720 to 960 hours of labor is actually rather significant. That's 720 hours closer to a permanent solution, assuming the researcher hired is a good one. And hey, you'd be indirectly employing someone for a long period of time.

Even if it is spent on hardware, most research servers are around $100k to $150k. $30k would put a significant dent in that allowing them to run algorithms on site.


Just remember that most (>99%) research computing is still performed on supercomputers. Never underestimate the usefulness of owning resources you are dependant upon.

 

[Atlas Folder]

To answer some of the questions I've seen:

I'm using AMD Phenom Quad-Core 9550 processors in MSI K9A2 Platinum motherboards. OS is XP Pro and RAM is 2GB per machine. There are 5 machines like that currently for 20 cards. The head or "server" of the rack that houses the data has one card. I have two cards in my home computer. 23 GTX295 total. All of the machines run Quad-Core SMP also. As a matter of fact you can view the rack's FahMon here.

The machines run fine on 2GB, the most I've ever seen one using is 1.35GB.

Jason

 

[h3llb3nd4]

a dead GPU??

 

[theeldest]

i didn't hear any results like a cure or something and since 2000....

i'm skeptikal about this f@h do you know exactly what data is processed ?

Stanford say is protein folding and other molecular dynamics but is really this?

It's actually processing the entire internet to find the best nudie pics.

 

[h3llb3nd4]

perverts wish that their pc could do that...

 

[theeldest]

Hey Jason,

I appreciate the Ayn Rand references.

 

[Atlas Folder]

Sure, GPUs die. Folding@Home is a pretty demanding application and I'm running them 24/7.

I went and checked the total current draw from the rack at the moment through my APC metered PDU (networked power strip). 46.4 Amps continuous across the whole machine.

46.6 * 120V = 5,592W

..and who is the Ann Rand person?

Jason

 

[mmaakk]

It's actually processing the entire internet to find the best nudie pics.

Yes and if you want a share of the results, WILL HAVE TO FOLD WITH US.