Nvidia GeForce 8800 capable of folding

Posted at: 12:00am 17th November 2006 by Ben Hardwidge

It looks as though Nvidia could soon be joining its nemesis-in-red ATi in the Folding@home arena, as a part of its GPU computing initiative.

Nvidia's Product PR manager for Northern Europe, Adam Foat, told Custom PC that 'We have kept many research groups, including Stanford [the university that runs the Folding@home project], briefed on the [G80] technology as we completed the software and hardware. Throughout next year, you'll see quite a bit of work from many academic researchers.'

In fact, it turns out that the green giant of graphics has been interested in GPU folding for a long time. We spoke to Vijay Pande, director of the Folding@home Distributed Computing Project, about the potential for folding on Nvidia GPUs, and he told us that 'Our GPU code was originally developed for Nvidia GPUs, but recent ATi GPUs (X1900 class) have really pulled away from the previous generation of Nvidia GPUs.'

However, he then added that 'The G80 may change that and we are looking into it. Our code is written in Brook, so it's not limited in any way to ATi GPUs, although there may need to be code changes to achieve optimal performance on other hardware such as the G80.'

ATi has already proved that a GPU can easily rival top-end CPUs when it comes to folding. We recently ran the GPU folding client on a Radeon X1950XTX, which managed to produce a work unit worth 330 points in just 15 hours.

Folding is just one of many parallel computing tasks that Nvidia is looking at processing on its GPUs. Foat expanded on the move into GPU-processing by saying that, 'The main focus of the computing effort at Nvidia is commercial applications that move applications from clusters of computers to one to four GPUs.

'Applications such as medical signal processing from diagnostic equipment, electromagnetic simulation, bioinfomatic searches or database analysis are all good fits for GPU Computing. The range of applications is actually quite wide. We're not trying to simply replace CPUs in high performance applications. Where a CPU is focused on running many different applications, the GPU can process immense amounts of data through many individual thread processors on the GPU and through a very wide, high-speed memory system.'

Folding@home is a distributed computing project that simulates how proteins fold for the purposes of medical research. By getting thousands of people to donate their spare CPU cycles to the project, Stanford university hopes to help find cures for Alzheimer's, Huntington's, Parkinson's disease, and many Cancers and cancer-related syndromes. You can help the project by downloading the client from here, and you can join Custom PC's team by using the id 35947.