Overclock.net banner

[DVHARDWARE]Researchers create supercomputer with four GeForce 9800 GX2 cards

6K views 49 replies 35 participants last post by  Fearless 
#1 ·
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 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 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:

# 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

The medical researchers ran some benchmarks and found that in some cases their 4000EUR desktop superPC outperforms CalcUA, a supercomputer with 512 AMD Opteron cores 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.

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:

Link

Code:
Code:
http://www.dvhardware.net/article27538.html
 
See less See more
3
#9 ·
Quote:

Originally Posted by 003 View Post
I thought that

A) CUDA does not work with SLI or multi GPUs
B) SLI does not work on AMD motherboards
C) 9800GX2 only scales to two cards, and nvidia does not support anything over 4 GPUs in parallel

How does this work?
All of this only goes for SLI. You can have as many GPUs as the OS allows, using them for gaming depends on drivers, chipset, and so on, and that's where the limitation lies. SLI and Crossfire.

These cards are not running in SLI.
 
#10 ·
Quote:

Originally Posted by 003 View Post
I thought that

A) CUDA does not work with SLI or multi GPUs
B) SLI does not work on AMD motherboards
C) 9800GX2 only scales to two cards, and nvidia does not support anything over 4 GPUs in parallel

How does this work?
My guess is the program/software they are using indivually assigns work to each of the gpu cores. Dont think about this application in the same terms as you think of a normal gaming pc running a game.
 
#13 ·
Quote:

Originally Posted by Slaveofthebeast View Post
You guys are thinking about this all wrong. The programs they're using are not going through SLI or Gaming drivers to use the gx2's, they're assigning huge calculations to each one, to solve a larger problem. Using gpu's like this has enormous potential .
+rep
Thank you...It's not hard people.
Gpu's are designed to crunch in this way.
Obviously this is better than 200 cpu's, think about it.
If you try to render graphics on a cpu, it's hundreds of times slower than on a powerful gpu, because of the way it's designed.

I wish they would find more ways to implement gpu's this way, like i said in the other thread concerning CuDA
 
#15 ·
Quote:

Originally Posted by dumont View Post
Check out the monitor in the first picture, it looks like WinXP from the bar at bottom and top blue frames.

Quote:
Here's a look at the specifications of the FASTRA desktop superPC:

# 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
Yeah?
 
#21 ·
It would be easier to design a separate multi-flatbed platform for multiple GPU beds, stacked, and watercooled.
Would cost less, operate more efficiently, and take up less space.

Have to wonder why Intel's and AMD's CPUs are used in Supercomputers instead of GPUs. Well at least some of the Supercomputers, after all, calculating Physics is a big part of supercomputer's life, why not let specialized GPU deal with it much more efficiently.
 
#22 ·
Quote:


Originally Posted by dumont
View Post

It would be easier to design a separate multi-flatbed platform for multiple GPU beds, stacked, and watercooled.
Would cost less, operate more efficiently, and take up less space.

Have to wonder why Intel's and AMD's CPUs are used in Supercomputers instead of GPUs. Well at least some of the Supercomputers, after all, calculating Physics is a big part of supercomputer's life, why not let specialized GPU deal with it much more efficiently.

Nvidia sells Quadro and Tesla rackmounts and such, but they are much more expensive than retail cards even though they have the same hardware.
 
#23 ·
Quote:


Originally Posted by savagebunny
View Post

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 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 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:

# 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

The medical researchers ran some benchmarks and found that in some cases their 4000EUR desktop superPC outperforms CalcUA, a supercomputer with 512 AMD Opteron cores 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.

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:

Link

Code:
Code:
http://www.dvhardware.net/article27538.html
a ting of beauty. if they are willing to sell one of those i call dibs!! LOL
 
#26 ·
Quote:


Originally Posted by 003
View Post

I thought that

A) CUDA does not work with SLI or multi GPUs
B) SLI does not work on AMD motherboards
C) 9800GX2 only scales to two cards, and nvidia does not support anything over 4 GPUs in parallel

How does this work?

the cards are not linked in QUAD-SLI, so you can use them on an AMD chipset. and i believe they are using some other program to use all 8.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top