Monday was a busy day in the battle between Nvidia and AMD (the parent company of ATI), as both companies unveiled powerful new graphics processors. However, what’s interesting is the change of direction that AMD is taking when it comes to the market.

First, there’s Nvidia, which launched its new GeForce GTX 200 GPUs. This is a family of processors that represents the most powerful from Nvidia yet. This performance will not come inexpensively, though. Graphics cards built on the high-end GTX 280 chip are retailing for around $650, while graphics cards built on the slightly less powerful GTX 260 chip are retailing for about $400.

AMD held a press conference in San Francisco Monday morning where it unveiled its new R700 chip architecture. This is a radical new design from AMD, and the company says that it’s taking a different approach to the market. Instead of creating a bigger, hotter GPU that appeals to hardcore gamers and then creating smaller, less expensive versions of that chip for more mainstream markets, AMD wants to create highly scalable chips yet powerful chips that can serve the needs of everyone immediately.

On paper, the R700 is an impressive chip. Though about the size of a fingernail, AMD says the R700 can deliver slightly over a teraflop of computing power. A teraflop is one trillion calculations per second. The R700’s small size (roughly 16mm on a side) means that the R700 uses about 110 watts of power and can appear in single-slot solutions without the need of a secondary power connection. "This is a solution for everybody," said Rick Bergman, the general manager of the graphics product group at AMD.

Mainstream Radeon HD 4850 cards based on the R700 chip will begin appearing on shelves next week for approximately $200. The Radeon HD 4870, a more powerful card designed for enthusiasts, will ship July 8. And an ultra enthusiast version, which will feature two R700 chips in it, will ship in two months. AMD says that the Radeon HD 4870 and the yet-to-be-named ultra enthusiast version will be dual-slot solutions. Bergman also said that AMD is working on a system that supports up to four R700 chips, which he said is almost five teraflops of graphics computing power.

In a question-and-answer session, Bergman said that the R700 will turn the tables in AMD’s battle with Nvidia. He cited the chip’s high performance, low power, and wealth of features. For example, he said the R700 supports Microsoft’s new DirectX 10.1 API, whereas the new Nvidia chip only supports DirectX 10. Also, the Radeon 4000 series will use new GDDR5 memory, whereas the new Nvidia chip still uses GDDR3. And by engineering a small but efficient chip, AMD has the advantages of lower costs. "Our die is about the same area as a dime. [Nvidia’s] is about the same area of a quarter. Despite that, we’re going to be 80 to 90-percent of their performance," Bergman said.

Source: IGN