• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Tesla PERSONAL supercomputer

C

crazjayz

2[H]4U
Joined
Jul 31, 2005
Messages
2,093
This morning in Austin, TX, Nvidia released information on their newest model in their Tesla supercomputing lineup, the GPU-based Tesla Personal Supercomputer. It's based on the Tesla C1060 GPU Computing Processor utilizing the NVIDIA® CUDA™ parallel computing architecture. Supposedly, it's capable of +2 teraflops of computational processes. Velocity Micro has their ProMagix™ VSC240 Workstation, which can processes 4 teraflops of calculations with its 4x C1060 GPUs and 2x Xeon processors.

Now for the completely pointless question. How many ppd on this bad boy :D
 
1 Tesla C1060 = GTX 280 minus DVI ports so you can guess the PPD. 4 Teraflops = 32000 ppd (34000-36000 with the dual quad Xeons folding as well).

 
Has anyone actually folded on a C1060? I realize that the specs (GPU speed wise) are the same as a GTX280, but iono, I just feel that it should be "optimized" for CUDA and whatnot, thus yielding much higher PPD. Else, why would they even market this card. CUDA computations don't require 4GB of video RAM, least I don't think. I feel that more companies/whomever uses this type of card would just buy GTX280s or 9800GX2s instead. I'm just wondering what the benefit is. And I'm sorry for the noob-ish question. :(
 
The reason for the "professional" grade cards are two-fold, first better direct support from nvidia and second support for the cards in cad/3d apps.
 
crazjayz said:
Has anyone actually folded on a C1060? I realize that the specs (GPU speed wise) are the same as a GTX280, but iono, I just feel that it should be "optimized" for CUDA and whatnot, thus yielding much higher PPD. Else, why would they even market this card. CUDA computations don't require 4GB of video RAM, least I don't think. I feel that more companies/whomever uses this type of card would just buy GTX280s or 9800GX2s instead. I'm just wondering what the benefit is. And I'm sorry for the noob-ish question. :(
Click to expand...

This isn’t so much for companies as it is for scientific study of all kinds.

Stanford uses an existing technology for its work and that technology is shared throughout various scientific communities. Astrophysics, Astronomy and virtually all the Bio Sciences and Earth Sciences use the same basic programs written in FORTRAN and the programs they use are easily transferable between the sciences. Desktop computers were never really adapted to this form of computing.

Back in the day we had the “Cray” super computer:
http://www.computerhistory.org/brochures/companies.php?alpha=a-c&company=com-42b9d5d68b216#

They were built in pie like slices and to get a full unit was a multimillion dollar investment and that didn’t include the terminals to access the computer. Things got a bit smaller and no longer required liquid nitrogen cooling:

http://www.cray.com/Home.aspx

So, today we have essentially the equivalent to a Cray or more on our home desktops. Strange huh?

Now, we still don’t have enough power so clustering, or more properly Distributed Computing is the big thing. Keep in mind, there will NEVER be enough power thus the process keeps growing and growing.

Nobody has ever been able to build a single computer big enough to do what we do for Stanford. Even IBM’s super computers are used in clusters to process billions of terabytes of data which still isn’t enough.

The CERN project, if it ever gets off the ground and doesn’t create a black hole that eats the world, will produce enough data in a few hours to fill every hard drive in the world.

http://public.web.cern.ch/public/

The answer of course is simple “More “D””;)
 
There are a few of them in the NV booth here at SC08. This thing even looks mean.

crazjayz, The C1060 has the exact same core as the 280, so at the same clock speed it should give the same ppd. There are no CUDA "optimizations" afaik.

There are several reasons to use Tesla if you're computing over the GeForce line. Yes, the additional memory is one big reason. FaH may not need it, but plenty of CUDA apps do. Another is the additional testing done on the cards before they leave the factory.
 
bassman said:
There are a few of them in the NV booth here at SC08. This thing even looks mean.

crazjayz, The C1060 has the exact same core as the 280, so at the same clock speed it should give the same ppd. There are no CUDA "optimizations" afaik.

There are several reasons to use Tesla if you're computing over the GeForce line. Yes, the additional memory is one big reason. FaH may not need it, but plenty of CUDA apps do. Another is the additional testing done on the cards before they leave the factory.
Click to expand...

Another reason is the lack of dvi ports so no more issues with extending the desktop and doing voodoo stuff to make it run, The drivers might also be different in that aspect as well.

 
This thing does seem pretty cool. I would take one. I don't really do anything other than folding that would use that ability much, but it would still be cool.

 
BillR said:
This isn’t so much for companies as it is for scientific study of all kinds.

Stanford uses an existing technology for its work and that technology is shared throughout various scientific communities. Astrophysics, Astronomy and virtually all the Bio Sciences and Earth Sciences use the same basic programs written in FORTRAN and the programs they use are easily transferable between the sciences. Desktop computers were never really adapted to this form of computing.

Back in the day we had the “Cray” super computer:
http://www.computerhistory.org/brochures/companies.php?alpha=a-c&company=com-42b9d5d68b216#

They were built in pie like slices and to get a full unit was a multimillion dollar investment and that didn’t include the terminals to access the computer. Things got a bit smaller and no longer required liquid nitrogen cooling:

http://www.cray.com/Home.aspx

So, today we have essentially the equivalent to a Cray or more on our home desktops. Strange huh?

Now, we still don’t have enough power so clustering, or more properly Distributed Computing is the big thing. Keep in mind, there will NEVER be enough power thus the process keeps growing and growing.

Nobody has ever been able to build a single computer big enough to do what we do for Stanford. Even IBM’s super computers are used in clusters to process billions of terabytes of data which still isn’t enough.

The CERN project, if it ever gets off the ground and doesn’t create a black hole that eats the world, will produce enough data in a few hours to fill every hard drive in the world.

http://public.web.cern.ch/public/

The answer of course is simple “More “D””;)
Click to expand...


Holy Crap! Someone else old enough to remember when Cray was the powerhouse to have. I recall United Airlines building a 2story paorking lot to house the mainframes downstairs - only to be replaced by 4-Crays.:D



 
Wheresatom said:
This thing does seem pretty cool. I would take one. I don't really do anything other than folding that would use that ability much, but it would still be cool.

Click to expand...

They were giving away one a day at SC08. <crosses fingers>
 
AMD_Gamer said:
thanks a lot that would be awesome
Click to expand...

My cell phone snapshots were supremely crappy. LOL

pic-0002_1.jpg


pic-0001_1.jpg


NVIDIA was taping some promotional video for the thing yesterday; here's the youtube link with some better pictures:

http://www.youtube.com/watch?v=sUGUCaxrrL0
 
You must log in or register to reply here.
Back
Top