1 Votes | Average: 5 out of 51 Votes | Average: 5 out of 51 Votes | Average: 5 out of 51 Votes | Average: 5 out of 51 Votes | Average: 5 out of 5 (1 votes, average: 5 out of 5) Loading ... Loading ...

by Mike Zazaian September 29, 2006 - 5:33pm, No Comments

A prototype of ATI's upcoming FireStream GPGPU card

While the raw power of GPUs have only been harnessed for graphics applications to this point, ATI’s announcement of stream computing today will allow users to harness much-needed GPU muscle for non-graphics applications.

After only hinting at the notion in a press release last week, ATI officially announced today the integration of Stream computing into future graphics card models. Stream computing marks the first time that GPGPUs, or general purpose graphics processing units, will be available to commercial enterprises. ATI had opened up its hardware and programming interfaces to Universities for research purposes, but aside from that the GPU had been on lockdown to consumers. Now GPGPU-enabled video cards from ATI will allow users to harness the multiple threads of GPU throughput for a vast array of processor-intensive applications. Said ATI CEO Dave Orton:

The raw computational power of these DirectX-based processors is incredible and harnessing those capabilities for general purpose applications makes a lot of sense. They were designed with more than just graphics in mind, and today we’re seeing the initial fruits of those labors with the introduction of our stream computing initiatives.

Stream computing will allow the use of GPU processing in tandem with conventional CPUs to achieve enormous amounts pure data processing. Whereas CPUs usually have only 4, 8 or 16 threads for process execution, GPUs contain as many as 48 threads, allowing GPUs to handle several times more operations per second than CPU-only processing.

Stream computing will be available on ATI’s Radeon X1900 graphics card, as well as a number of other cards currently in development. Immediate commercial applications include financial forecasting, database searching, consumer applications, graphics, and scientific research, where projects such as Folding@Home are expected to gain a 40 fold increase in overall research productivity.

[ATI Press Release]