Why Nvidia's CUDA might change the face of high speed computing

Why Nvidia's CUDA might change the face of high speed computing

The GPU battle is being fought on far more fronts than consumer desktops. While the new Fermi cards may look hideously impractical for your average home rig, the applications it has for science are more significant.

From our recent results, its easy to think that AMD has won another round in the Red vs Green graphics battle. You could also start to wonder whether this could spell the final round for Nvidia. But if you look at the bigger picture, there's more to it than framerates.

Biologists, in particular, often need massive computing power for rendering 3D protein structures, and being able to do their work on a desktop workstation speeds up their work by months, if not years, as they save time that would otherwise be spent waiting for a slot on a supercomputer.

The advantages of Fermi, and of Nvidia's approach in general, is that it's looking at the GPU almost as a CPU - something that can take thousands of simple calculations and run them in parallel in a different way to the CPU's usual considered, measured approach to instructions.

The CUDA programming capability is part of Nvidia's push to gain space in high performance computing, and the benefits it adds to gaming are almost secondary in importance at this point.

You have only to look at the new Adobe CS5, with CUDA and PhysX optimisations, to realise that more companies than just Nvidia are taking sides in the graphics war.

Will Adobe's punt look shortsighted in a year's time? It's hard to tell. Although Nvidia says that its CUDA system can run on AMD's cards, it currently doesn't.

And if Adobe takes a punt on Nvidia over AMD/ATI, others will follow: High end Mac Pro workstations are generally used for design, and having CS5 optimised graphics cards in those systems will be essential.

If Apple puts Nvidia cards in Mac Pro to satisfy designers' needs, there will be pressure to invest in Nvidia's technology in the rest of their range - as much as anyone can pressure Apple, that is.

Can Nvidia claim enough of a niche to keep developing its technology into the powerhouse it clearly wants it to be? Yes, but it will have to keep satisfying the demands of both scientists, designers and gamers, in order to do it, and right now, they only have two out of three cornerstones to success.

See more about:  cuda  |  nvidia  |  amd  |  intel  |  cpu  |  adobe  |  cs5  |  gpu  |  ati  |  graphics

Latest Comments

From our Partners

PC & Tech Authority Downloads