$5K PC Takes On $4.6M Supercomputer

Antwerp (Belgium) - Recent advances in general purpose GPU computing are beginning to shift perceptions in supercomputing applications. Belgian researchers have assembled a relatively simple enthusiast PC with an emphasis on graphics processing capability, which beats a multi-million dollar supercomputer in its target application.

The desktop PC, called Fastra, was built with a focus 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. As these 3D images can be quite large, advanced reconstruction techniques can sometimes require weeks of computation time on a regular PC. Which means that supercomputers are usually required to process computer tomography (CT) images.

While it is an impressive example how GPUs can be applied in non-traditional ways, there are a few notes to be added. Of course, GPUs cannot replace traditional supercomputers, which still can be applied to applications with a broader range. Also, supercomputers usually carry huge memories, often in the Terabyte range, which cannot be matched by today’s GPU clusters. When we talk to scientists working with supercomputers and GPUs, they typically believe that future supercomputers will not completely transition to GPU clusters, but may develop into systems that consist of a traditional supercomputer structure as well as GPU capability.

An interesting side note about the Fastra PC is its motherboard. Eagle-eyed readers may have noticed that the MSI K9A2 Platinum board is not an Nvidia SLI-based board, but uses AMD Crossfire (780 chipset). The simple reason to choose this board may have been cost, but it is unlikely to impact the performance of the system : CUDA does not support SLI at this time, which means that the GPUs have to communicate with each other as well as with the CPU via PCI Express. The researchers claim that they have not seen any impact on performance and the GPUs apparently are scaling well.

Wolfgang Gruener
Contributor

Wolfgang Gruener is an experienced professional in digital strategy and content, specializing in web strategy, content architecture, user experience, and applying AI in content operations within the insurtech industry. His previous roles include Director, Digital Strategy and Content Experience at American Eagle, Managing Editor at TG Daily, and contributing to publications like Tom's Guide and Tom's Hardware.

  • Blackopsninja
    sweet
    Reply
  • dittopb
    I think the reason why they choose MSI K9A2 Platinum board is the PCIe slot spacing. I checked Asus and Gigabyte boards with quad PCIe 2.0 support and found to be unevenly spaced and would be hard to fit 4 dual-slotted GPUs.
    Reply
  • hixbot
    Ok, now we know how well 8 GPU cores handle supercomputer-tasks. Lets see how the $4.6M supercomputer can handle Crysis!
    Reply
  • hughyhunter
    It's about time someone realizes we build for shear power!!!
    Reply
  • lopopo
    It's about time someone other than folding did this. Only hope others will do the same.
    Reply
  • dobby
    interesting choice of CPU, i know that this doesnt use the CPU directly for it Processing power, but you would think that they would go the full hog and use a skull trial system
    Reply
  • khaydin
    8GB of ram on Windows XP? Hopefully it's 64bit XP otherwise they're wasting more than half of that ram...
    Reply
  • customisbetter
    how does one, let alone 4 9800 gx2s get 12k in 3dmark?
    Reply
  • hughyhunter
    9080880 said:
    how does one, let alone 4 9800 gx2s get 12k in 3dmark?
    I agree... I'm interested to hear a reason why. Also... I didnt know there was support for four GX2's. How do you run 4 of them? Their is only 1 connector!
    Reply
  • khaydin
    hughyhunterI agree... I'm interested to hear a reason why. Also... I didnt know there was support for four GX2's. How do you run 4 of them? Their is only 1 connector!
    They're not setup in SLI since the motherboard and CUDA doesn't support SLI. So if they were doing 3dmark only 1 of those 4 would be getting benched.
    Reply