CSIRO's latest computer cluster gets its grunt from games

Technology designed to blast aliens in computer games is part of the latest GPU (graphic processing unit) computer cluster that will process CSIRO research data thousands of times faster and more efficiently than a desktop PC could.

GPUs aren't just useful for image data but are good for many big science challenges processing petabytes of data and more very quickly. Speeds of 30 to 70 times faster than CPUs are common and using GPUs to analyse complex research data is becoming a global trend in computing.

CSIRO's GPU cluster will be used for research in advanced materials, cloud computing, data and visualisation tools, genetics and more, including:

- figuring out where tiny fragments of genetic code sit on a genome
- 3D reconstruction of medical images from the Australian Synchrotron
- modelling the interactions between nutrients and plankton in the oceans.

http://www.csiro.au/resources/GPU-cluster.html

Category:

Science & Technology

Tags:

• CSIRO
• research
• computer
• gaming
• gpu
• supercomputer
• computer_science
• data_visualisation
• modelling
• computer_imaging
• computer_cluster

[boygenious92]

is there any software that i can get or stuff i can order or manuals any thing i want to build a small one at home for video rendering and my phenom 2 is good but not good enuff so if you could give me some information or some thing to get me started id be grateful thanks

[csiro]

@boygenious92 our response to your question is quite long, so we'll have to give it to you in a few sections...

[csiro]

The GPU cluster is based on a standard cluster of PCs (also called a Beowulf cluster), constructed by connecting many low cost PCs or workstations together using an isolated/private network to form a "super computer" (see ‘computer cluster’ and ‘Beowulf computing’ on Wikipedia). There are lots of online and print resources on the topic, and clusters can be built using Linux (see the ‘Beginners’ page under ‘HPC Resources’ on the linuxhpc website) or Windows HPC Server operating systems.

[csiro]

In our cluster, each workstation has two GPUs installed on it. We use Message Passing Interface (MPI) software to code parallel programs across the cluster, and NVIDIA CUDA software or the Khronos OpenCL standard to code portions of these programs on the GPUs.

[csiro]

Some scientific, 3D rendering/animation, or engineering software already comes with the ability to run in parallel on a cluster and/or on GPUs, these types of programs can be used without needing to familiarise yourself with computer programming, MPI, or CUDA/OpenCL. We suggest you check out an online video rendering user group or community - they should be able to help with GPU or cluster enabled software options specific to what you want to do.

[csiro]

Hope that's helpful!

[davidmthekidd]

Yea... But can it play crysis!

[Devorvan]

Will it blend?

[Storhonta]

Can't you just download a benchmark for computers and just see how good the thing really is ;P

[csiro]

@boygenious92 Hi, we run both Linux and Windows HPC 2008.