Nvidia Launches Super New Processor

Arthur Cole

The age of the "personal supercomputer" may be at hand. Nvidia announced this week a line of GPU-based processor boards said to vastly boost the performance of everything from single PCs to large-scale server clusters.

 

The new Tesla family of GPUs is available in three configurations. There's the Tesla GPU Computer Processor board that offers 128 parallel processors capable of a combined 518 gigaflops and can be installed in existing systems outfitted with high-performance CPUs.

 

The Deskside Supercomputer is a scalable system that ties two Tesla GPUs to a PC or workstation via PCI-Express. Several Desksides can ramp up the typical desktop to the 8-teraflop level.

 

Finally, there is the Computer Server, a 1U design that holds eight Tesla GPUs to deliver more than 1,000 parallel processors to bring multiple teraflops to the typical cluster.

 

Applications can be developed for the Tesla via Nvidia's CUDA language, currently supported by Linux and XP. The C-language environment is geared toward programs that cater to parallel processing.


 

At the moment, supercomputing capabilities like these are beyond the scope of all but the largest enterprises or those dedicated to data-intensive applications like medical research and geoscience. But as history shows, these top-end systems have a way of trickling down to more esoteric levels sooner rather than later.



Add Comment      Leave a comment on this blog post

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

null
null

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.