Make your own free website on

It was one year ago that Nvidia announced the GeForce2 Go, the company's first mobile graphics chip. Combining the solid 3D performance of the GeForce2 MX design with power-saving technology, the GeForce2 Go has helped make 3D performance a competitive point between notebooks since it shipped in systems last spring. Now Nvidia has a new mobile design currently known by its internal code name, the NV17M. The new design is a significant step in 3D gaming performance and in DVD acceleration.

The demands placed on mobile graphics chips are different from those placed on their desktop counterparts. Notebook makers, always eager to increase battery life, require all components to consume as little power as possible. Nvidia's PowerMizer technology for the NV17M uses all the conventional tricks--shutting down unused parts of the chip and varying the chip's clock speed and voltage depending on application demand--to keep power consumption down in the range of .5 to 2 watts. But the biggest potential for power savings comes in offloading tasks from the CPU for overall system power savings. In particular, Nvidia has added more DVD acceleration features to the NV17M, taking over practically all decoding takes from the CPU except audio processing. This should make it much easier to watch DVDs on the go without worrying about running out of battery power.

For better 3D performance, Nvidia has boosted the clock speed of the .15-micron chip to 250MHz for both the chip core and memory. In the high-end configuration, this memory speed will provide 8GB per second of memory bandwidth--just as much as a GeForce3 Ti 500. However, the chip still has the dual-pipeline design of the GeForce2 Go and GeForce2 MX, as opposed to the four pixel pipelines of the GeForce3 and GeForce2 GTS. This means that--in contrast with most recent graphics designs--the NV17M may actually have a surplus of memory bandwidth in its high-end configuration. Perhaps the biggest advantage of this jump in performance is that it allows for the sort of high-resolution antialiasing introduced with the GeForce3. Not only does the NV17M do Quincunx multisample antialiasing, but it also integrates much of the GeForce3's memory architecture improvements, resulting in what Nvidia claims is a fivefold increase in antialiasing performance over the GeForce2 Go.

There are two basic configurations, the 64-bit DDR 2xx series and the 128-bit DDR 4xx series. These options put 32MB or 64MB of memory on the same package as the chip, keeping the interface to the notebook motherboard consistent. High-bandwidth 128-bit designs typically take up quite a bit of board space by doubling the aluminum traces that connect the memory to the chip, but Nvidia has managed to keep the NV17M small even in this case, putting the chip on the bottom of the mobile package and the memory on the top.

Expect to see NV17M-equipped notebooks from Toshiba and other Nvidia partners next spring. It's also in about that time frame that we'll start seeing ATI's next mobile chip, the Mobility Radeon 7500, ship in systems.

Posted on November 16th, 2001

<< Back