AMD, Business, Companies, Graphics, Hardware, Intel

UPDATED: Nvidia set to launch Tesla powered Personal Supercomputer

Today is the first day of SuperComputing 08 conference held in Austin, Texas. A lot of companies are bringing out the big guns for that one, and one of the companies that could have the largest one is Nvidia and its partners.
Regardless of what you may think of CUDA, this API really took off in scientific community. Young enthusiasts started to build personal supercomputers, and Nvidia CUDA guys got the idea of creating a personal supercomputer when they saw FASTRA project from University of Antwerp. FASTRA is being used for computational topography, but many other universities are doing exactly the same.
Fast forward to SC08, and we have Nvidia launching its own product, Tesla Personal Supercomputer. In a nutshell, we’re talking about Intel or AMD powered system feeding four Tesla C1060 cards. We heard that the reference system uses two nForce 200 chips (even though SLI is not used), Intel Core i7 processor and 8GB of system memory. In a nutshell, this is what you get:
•    No less than a 4-core CPU
•    No less than 4GB of memory (expect 8GB in most systems)
•    No less than four Tesla C1060 boards
•    No less than 4 TFLOPS in single precision format and additional 0.56 TFLOPS in dual precision
The company assembled quite a lot of OEMs and system integrators, such as ASUS, Dell, Lenovo, Scan and many others – all in a bid to jump-start revenue coming from the Tesla product line-up. In a conference call for Q2’08 results, Jen-Hsun stated that Q2’09 will see impact from Tesla line-up, and my personal take is that this product could be a surprising success.


All of the images in gallery are 1000 pixels wide… if you would like to see wider or narrower galleries, it would be great if I could get some feedback and suggest a resolution 😉
The reason for it is quite simple: unlike companies that are advertising personal supercomputers and charging them tens of thousand dollars (Cray CX1 and Tyan PSC come to mind) – Nvidia is targeting a price of less than 9,999.99 USD. Yep, that’s correct, 4+ TFLOPS supercomputer for less than 10 grand.
At the time of writing, I haven’t been able to found are those Tesla cards using older 65nm or new, power-saving 55nm chips. If we would judge by chips that power new Quadros, 55nm should be under the hood, but then again, Nvidia has a large nVentory of older 65nm chips, so the company might be inclined to clean them out.
Somehow, I would bet that this is not the only supercomputer announcement from companies in Red-Green-Blue area. After all, certain motherboards from ASUS and GigaByte could fit six single-slot cards… can you say FireStream and 6 TFLOPS?
To end this article, here is some food for thought: Nvidia got inspiration to create this system after seeing tens of enthusiasts build their own supercomputers. 11 years ago, overclockers and gamers jump-started the IT industry we know today by overclocking and modding their own systems.
Can this underground movement turn desktop computers into handhelds and supercomputers onto desktops? Only time will tell. After all, what CPU, GPU or motherboard manufacturer does not encourage overclocking or offer an overclocking/gaming product line? After all, all of these products are based on overclocking Celeron 300A to 450 MHz, Dual-Celeron Abit BP6 and so on. Makes me wonder who will remember those crazy doctors that created FASTRA 10 years from now.

UPDATE Nov 18, 01:23AM CET: For some odd reason, gallery didn’t work – uploaded the pictures again. It should work now.

  • tapper

    imagine one of thos babys for folding at home lol then thers also the electric bill

  • fastra produce about 48 000 ppd …

    🙂

  • Well, with all due respect to Folding@Home project, I view this as a much bigger picture… the time has come for professional workstations to have more than enough horsepower. We are in game changing situation.

    Several weeks ago, I was in HQ of a certain legendary automotive company. Most people I talked there spoke about the need to increase the amount of computational power on desktops, instead of sending their projects to process on clusters. I believe 4TFLOPS is just the start, but if Nvidia’s or AMD’s partners would start to offer 10TFLOPS machines (for instance, 6xGT212 chip or 6xRV790), you would be in for a quantum shift of workstation market, especially in automotive, airline or any other CFD-dependent industry.

    After all, even Intel’s Larrabee is debuting in workstation market to capture as much market as possible.

  • Boris, when it comes to Fastra, bear in mind that it has a quad-core CPU from AMD. In order to get high GPGPU performance, regardless of that being Folding@Home or Dassault’s CFD simulation, you need to have one CPU core dedicated to feeding the GPU.

    I hope that in future, Nvidia will start putting the Tegra core onto Tesla cards, so that card itself can access the hard drives and system memory, thus releaving duties from the CPU itself. Only time will tell 😉

    P.S. My Phenom system can choke when I run F@H on 9800GX2 + GTX280 (4 CPUs taking care of 3 GPUs), yet alone taking care of four 9800gx2 boards.

  • Looking at the pictures you posted, I wonder how noisy those systems must be, and considering they are workstations and not servers, they are usually inder the desk of the user. The cooling must be a real challenge, on those, but you and I know there’s a solution to both problems, don’t we? Too bad nobody has really had the guts to invest REAL money on this.

    Garacs1

  • It is weird though that nVida targets this only towards the scientific community – how about all those working in heavy graphic applications? With video files approaching the gigabyte per second level, a Supercomputer like this would be totally revolutionary for the post workflow of smaller film/video production houses.

    Exciting times, indeed.

  • Well, when it comes to video, Quadro FX is already used extensively in broadcast. Personally, I am using Skulltrail machine with GTX280 (for gaming+CS4) and Quadro FX 4600 SDI (for video capture).

    And instead of introducing tesla, I would really like to have CUDA-enabled drivers for FX4600SDI, since the last drivers came out half a year ago.

    Tesla will be targeting scientific community, Quadros are for us, visual types 😉