Chip-maker Nvidia Corp. unveiled two graphics microprocessors that offer twice the horsepower of previous models and what the company touts as a design milestone: a multitasking brain capable of efficiently handling a variety of non-graphics calculations.
The Santa Clara-based company said the target market for its GeForce 8800 graphics processing units, or GPUs, will initially be hardcore gamers looking for improved graphics and realism. The graphics cards that are powered by the new chips will sell for $599 and $449.
The new chips are based on a revamped architecture that makes it easier for programmers to focus on artwork and textures instead of tackling programming bottlenecks, the company said.
They also boast twice the number of transistors — 681 million per chip.
The company is also touting a design innovation that allows the chip to render images while performing complex non-graphic calculations, a feature the company said will appeal to financial institutions and scientific researchers who need to process massive amounts of data. Customers, however, would need a workstation or other advanced setup to run such non-graphics calculations.
"This is opening up beyond gaming markets to a whole different level of computing," said Michael Hara, vice president of investor relations and communications.
Analysts said the new design signals a shift in graphics chips toward more high-level general processing, and ratchets up Nvidia's competition with rival graphics chip designer ATI Technologies, which offers similar processors and was recently acquired by Advanced Micro Devices Inc.
Both companies also are competing fiercely to ramp up graphics capabilities as consumers demand better video and images in products ranging from computer games to handheld digital music players.
Dean McCarron, an industry analyst at Mercury Research, said Nvidia's new chips could appeal to companies that rely heavily on graphics capabilities but also need specialized computing to process large data sets for tasks such as X-ray image rendering or analyzing geophysical data.
"Even competing against themselves it's a very significant improvement," he said. "Once this change happens there won't be any going back; we're going to see graphics chips being more general-purpose, much more adaptive devices."