Graphics chipmakers Nvidia and AMD today will release new graphics processing units claimed to take games playing, multimedia applications and perhaps serious supercomputing to new levels.
The chips from both companies are the first to offer up to a teraflop of very fast computing power for large-scale computing and amazingly vivid graphics. But the companies are taking radically different approaches, experts say.
Nvidia’s new teraflop chips are the GeForce GTX 280 and GeForce GTX 260 GPUs, which come with 240 and 192 cores, respectively.
Developers working on video coding technology say their power allows users to convert two hours of HD video in 35 minutes, compared to 5.5 hours for a 3GHz quad core CPU.
Nvidia said this opens a new opportunity for heterogeneous computing, in which the GPU can work hand-in-hand with the CPU, helping offload some of the big tasks of the CPU to improve overall performance in video game development, editing photos or even running a multimedia operating system like Windows Vista.
The Nvidia chips include 1.4 billion transistors running at speeds up to 1.6 GHz. “It’s one of the largest chips in mass production and, unlike some processors, it is not 60 percent cache,” said Andy Keane, GM of the GPU unit.
Meanwhile AMD, which acquired Nvidia’s top rival ATI in 2006, has its own teraflop graphics chip line, but takes a totally different approach, involving two chips linked on a board by a proprietary interconnect and delivering sizzling performance.
AMD says its 4850 device, at about 110W and costing US$199 will deliver about 75 percent of the performance of Nvidia’s high-end GTX280, which costs
$649 and dissipates 236W. But two of the AMD parts on a single board should run about 30 percent higher than the Nvidia device..
The AMD chips use more than 500 cores, more than double the 240 cores on the new Nvidia parts.
Rick Bergman, GM of AMD’s graphics division, said the company’s focus on a more mainstream design would enable it to roll out a version for notebook computers that consumes less than 70W early in the second half of this year.
“There’s no way this new Nvidia core will be in notebooks” by then, Bergman said.
Nvidia and AMD’s biggest rival is Intel, which is working its own next-generation graphics architecture dubbed Larrabee, based on many streamlined x86 cores, and architecture based on ray tracing instead of the traditional graphics rasterisation approach. Larrabee is expected to reach the marketplace in late 2009 or 2010.
