Researchers at the University of Antwerp in Belgium have created a new supercomputer with standard gaming hardware. The system uses four NVIDIA GeForce 9800 GX2 graphics cards, it costs less than 4000EUR to build and thanks to NVIDIA's CUDA technology it delivers roughly the same performance as a supercomputer cluster consisting of hundreds of PCs!
This new system is used by the ASTRA research group, part of the Vision Lab of the University of Antwerp, to develop new computational methods for tomography. The guys explain the eight NVIDIA GPUs deliver the same performance for their work as more than 300 Intel Core 2 Duo 2.4GHz processors. On a normal desktop PC their tomography tasks would take several weeks but on this NVIDIA-based "supercomputer" it only takes a couple of hours. The NVIDIA graphics cards do the job very efficiently and consume a lot less power than a real supercomputer cluster.
The research group ASTRA, part of the Vision Lab of the University of Antwerp, focuses on the development of new computational methods for tomography. Tomography is a technique used in medical scanners to create three-dimensional images of the internal organs of patients, based on a large number of X-ray photos that are acquired over a range of angles. ASTRA develops new reconstruction techniques that lead to better reconstruction quality than classical methods.
Although our reconstruction techniques are very powerful, they have an important drawback: they are quite slow. As the 3D images that we normally deal with can be rather large (typically 1024x1024x1024 volume elements, or more), advanced reconstruction methods can sometimes take weeks of computation time on a normal PC.
Here's a look at the specifications of the FASTRA desktop superPC. The main reason why they configured an AMD system is because they couldn't find a motherboard for the Intel platform that could fit four GeForce 9800 GX2 graphics cards. Another interesting note is that this system doesn't need SLI, their application uses the NVIDIA CUDA programming model which makes all eight GPUs work in parallel. The researchers say they don't need SLI during a reconstruction as every graphics card communicates directly with the CPU, no inter-GPU communication is needed.
AMD Phenom 9850 processor + Scythe Infinity CPU cooler
4x MSI 9800GX2 graphics card
4x 2GB Corsair Twinx DDR2 PC6400 memory
MSI K9A2 Platinum motherboard
Samsung Spinpoint F1 750GB HDD
ThermalTake Toughpower 1500W Modular PSU
Lian-Li PC-P80 Armorsuit case
Windows XP 64-bit
A quick note for our American readers, a system with similar specifications would cost you around $3,900 at Newegg. Most computer hardware costs a lot more in Europe than in the U.S.
The biggest problem of the system is that these four dual-GPU graphics cards are cramped together and generate quite a lot of heat. The FASTRA uses aircooling and with the sidepanel removed the GPUs run at 55 degrees Celsius in idle, 86 degrees Celsius under full load and 100 degrees Celsius under full load with the shaders 20% overclocked. They have to run the system with the left side panel removed as the graphics cards would otherwise overheat but they're looking for a solution for their heat problem.
The medical researchers ran some benchmarks and found that in some cases their 4000EUR desktop superPC outperforms CalcUA, a 256-node supercomputer with dual AMD Opteron 250 2.4GHz chips that cost the University of Antwerp 3.5 million euro in March 2005:
You can read more about the FASTRA GPU SuperPC project over here. The site contains lots of background information, info on the hardware they used, more photos and benchmarks. More videos that explain what the project is all about can be found at YouTube.
In the video below Dr. Joost Batenburg takes you to the ASTRA-lab where he shows what tomographical reconstructions are and what the role of FASTRA is:
Basically the ASTRA researchers have created a very affordable computer system that can perform their tomography work very efficiently. For specific applications that can be massively parallelized GPUs are much faster than CPUs, for the tomography work the NVIDIA-based FASTRA system outperform a real supercomputer that's much more expensive, takes up more room and uses a lot more power. The Vision Lab is now planning to build a cluster of such systems, which will allow for real-time reconstruction of large 3D volumes.
In the future we'll see many more applications that will take advantage of the computational power of GPUs to significantly speed up specific tasks. Later this summer NVIDIA will unveil new software that will make HD video transcoding much faster and the firm will also roll out a software update that will add AGEIA PhysX support to all GeForce 8/9 and GeForce GTX 200 series graphics cards. Another application that benefits greatly from GPUs is Folding@Home, ATI GPUs are already supported for quite a while and soon NVIDIA GPUs will also get folding support. Adobe is also playing with GPU acceleration for Photoshop but the firm says this is just an experimental feature that may not be present in the next version of Photoshop.
The comments are owned by the poster. We aren't responsible for their content.
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Friday, May 30 2008 @ 00:19:11 CEST
I believe GPGPU is going to revolutionize a lot of things. Supercomputers used to be way to expensive for many companies or research teams like the ASTRA guys but thanks to NVIDIA CUDA you can now make your own supercomputer rig for just a couple of thousand dollars!
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Friday, May 30 2008 @ 16:05:27 CEST
That's pretty smart! You took something that was valued to over 3,5 mil euro 3 years ago, and overdone it for 4000 euro. I read the article and was very amazed! Good job! :D
If you can just find a way to remove all that heat, it would be perfect.. perhaps place some sort of heat absorbing material between the cards?
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Friday, May 30 2008 @ 17:10:38 CEST
I love his hand and arm movements during his commentary. And did you see his arms? That dude probably gets laid all the time! GO SUPER COMPUTERS! I bet he knows where I am right now, and is going to super compute me into Bolivia.
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Friday, May 30 2008 @ 18:49:56 CEST
Great job guys. Now, if you can write a CUDA code that does single particle reconstruction from EM images (3D reconstruction of protein particles from 2D electron microscopy images) in FASTRA, you'll make a lot of scientists happy :)
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Saturday, May 31 2008 @ 03:14:55 CEST
Take a few of those millions of dollars saved and get some decent cooling for the GPUs and you will pick up ~15% additional clockspeed easily.Save the money and dont use Lian-Li cases, as sexy as they may be. Function over form for science. Distributed Computing should also be considered as many gamers run multi GPU systems that sit idle for long periods of time. I myself have several multi-GPU systems so it would be nice to have something to crunch while Im at work and comps sit idle. ( not a big F@H fan so far ).
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Saturday, May 31 2008 @ 23:58:30 CEST
Each communicates directly with the CPU? Wouldn't that make the computation slower than if they were communicating only amongst each other? Although I suppose SLI for that many cards is hard to do...
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Monday, June 02 2008 @ 07:05:37 CEST
This is not a supercomputer - what a misuse of the word, in order to mislead. It's just that they need to do 3D calculations for this tomography, so GPUs are going to be the best thing to use. It's just a niche application, using appropriate tools.
Refers to those computing systems (hardware, systems software, and applications software) that provide close to the best currently achievable sustained performance on demanding computational problems.
Computer with enormous processing capacity and built with several multiprocessors.
a large, expensive and fast computer, usually arranged so that it can perform the same operation on all the items in a vector at once; useful for intensive number-crunching programs such as weather forecasting or high-quality graphics, but not usually effective for general parallel programs.
"Conventional" supercomputers serve a nice of problems just as well. Admittedly, this niche is broader than for GPU computing, but even GPU's are capable of dealing with an increasingly broad range of problems (check out gpgpu.org). The boundary is at least getting more and more vague.
16 or 8 displays (Score: 0) by Anonymous on Wednesday, June 11 2008 @ 10:17:40 CEST
What about hooking this thing up to 16 displays that surround the gamer, so the gamer doesn't need to use any controller to see what's happening around him at any angle ?
But what I really wanted to know is how many gigaflops can you get out of those cards ?
And maybe you guys can talk to the people of flight gear, which is an open source flight simulator. Try to simulate a flight at the speed of sound, or maybe even the speed of light ? (http://www.flightgear.org/)
Re: University of Antwerp makes 4000EUR NVIDIA supercomputer (Score: 0) by Anonymous on Tuesday, July 29 2008 @ 01:17:00 CEST
I sure hope you've upgraded to GT200 gpus by now. 4x GTX280 will give you twice the power you got with your GX2's.
Other then that, ignore the ignorance of others. The world needs people like yourself to push the demands on manufactures, or they wouldn't be any advancements. I'm sure your contribution to CUDA will benefit more the just your own research in the long run. People just don't realize that without these type of projects, there wouldn't be need for advancements as no developer would bother creating software that can't be benefited from.
