TAG | Nvidia turns GPU chips into hi-perf number crunchers |
|
| by Stephen Withers | |
| Thursday, 21 June 2007 | |
|
Nvidia's new Tesla GPU range packages this capability into three products: The Tesla C870 GPU Computing Processor is a PCI Express x16 card that occupies two slots in a PC. Its 128 processor core can deliver up to 518 gigaflops (one flop is one floating point operation per second). The Tesla S870 GPU Computing Server is a 1U rack-mounting package containing four GPUs. The Tesla D870 Deskside Supercomputer is a mini-tower unit with two GPUs that connects to a host system via a PCI Express x16 or x8 adaptor card. Software support comes in the form of CUDA, a development system including a C compiler, debugger and libraries. It runs on Linux and Windows XP. Early users welcomed the technology. "Many of the molecular structures we analyze are so large that they can take weeks of processing time to run the calculations required for their physical simulation," said John Stone, senior research programmer at the University of Illinois Urbana-Champaign. "Nvidia's GPU computing technology has given us a 100-fold increase in some of our programs, and this is on desktop machines where previously we would have had to run these calculations to a cluster. Nvidia Tesla promises to take this forward with more flexible computing solutions." "Geophysical and seismic analysis has historically been a batch process that takes many, many hours," said Steve Briggs, vice president of systems integration at Headwave, a Houston-based geophysical visualisation company. "Without inventions like Tesla and CUDA GPU computing technology, discoveries like Jack Field, where a large oil reserve was detected 27,000 feet beneath the Gulf of Mexico, simply would never have been found. NVIDIA's announcement today is going to make discoveries of this and even larger magnitude more possible." While GPUs have previously been used for scientific calculations (eg, the Folding@Home project has a client for recent ATI GPUs), Nvidia says CUDA enables direct implementation of parallel computations in the C language using an API designed for general-purpose computation. The first Tesla products will ship in August.{moscomment}
Get stories like this delivered daily - FREE - subscribe now
|
Tags
| < Next story in category | Previous story in the category > |
|---|
