Super PC that Beats Super Computer

Posted: April 11, 2010 in Computer
Tags: ,

Is it possible to fit the computing power of a large supercomputer cluster in the tight space of a PC case? Looking at new computational methods for tomography—a technique used by medical scanners to create 3D images—University of Antwerp researchers have built a budget supercomputer using 13 Nvidia GTX 295 graphics cards. tomography code is developed using the NVIDIA CUDA framework, a C-like programming language that allows for efficient programming of the NVIDIA GPUs.

The University’s first desktop supercomputer, The FASTRA, had only 4 dual-GPU graphics cards (for 8 total GPUs). By comparison, the FASTRA II has a total of 13 graphical processing units (six NVIDIA GTX295 dual-GPU cards and a single GTX275 single-GPU card). All this gives the FASTRA II 12 teraflops of computing power (12 trillion floating-point operations per second). Most home computers are in the in the gigaflops (billions of floating-point calculation sper second) range, and the TOP 500’s fastest supercomputers have reached petaflops (one quadrillion per second) heights.
FASTRA II needs all this graphics horsepower to deliver increased resolution on the University’s three dimensional bone scanning procedures. While all this power many not be practical for a den or study, it’s sure to impress the neighbors.
The University’s researchers had to make some modifications to the components to create FASTRA II, including a custom case to hold all of the graphics cards above the motherboard, a custom BIOS, and a customized Linux kernel (since an ordinary 32-bit BIOS couldn’t handle all of the graphics cards simultaneously).
FASTRA II Specification :
  • Intel Core i7 920
  • 6 NVIDIA GTX295 dual-GPU+ GTX275 single-GPU (total 13 GPUs)
  • 6×2GB Corsair DDR3 1333
  • Lian-Li PC-P80 Armorsuit Case
  • ASUS P6T7 WS Motherboard
  • Samsung Spinpoint F3 1TB Harddrive
  • Thermaltake Toughpower 1500W + 3x Thermaltake PowerExpress 450W
  • Price : Less than $ 8073

CalcUA Super Computer

  • 256-node supercomputer @AMD Opteron 250 2.4GHz (total 512 cores)
  • Price : $ 4,6 Million

Here’s Benchmark result, see the differences :O

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.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s