NVIDIA nForce Computex 2001 Preview

Integrated Graphics Processor (IGP)

As the name suggests, the Integrated Graphics Processor (IGP) provides all the functions of a standard north bridge, including CPU interface, memory controller, and AGP interface, but also adds an integrated graphics core. That graphics core is very closely based on the NV11 core used in the GeForce2 MX, making it far more powerful than any other integrated graphics chipset on the market today or on the horizon.

Like the GeForce2 MX, the nForce graphics core is clocked at 175MHz and features two pipelines, giving it a theoretical fillrate of 350 megapixels/sec. That also means we get a complete second generation transform and lighting engine. The kicker is that the graphics core gets an internal AGP 6X interface to the rest of the north bridge logic which could help tremendously in transferring things like triangle setup from the CPU to the graphics core's T&L engine or when the reserved frame buffer memory is used up and textures need to be swapped.

Of course, the biggest bottleneck in recent times for 3D graphics accelerators is memory bandwidth, which of course brings us to one of the most exciting features of the IGP - TwinBank Memory Architecture, which we'll talk about shortly. The combination of an integrated GeForce2 MX graphics core with TwinBank Architecture should put gaming performance in the neighborhood of a current DDR solution with an external AGP GeForce2 MX installed.

Remember that the integrated graphics on the nForce uses system memory, so 32MB of main memory is reserved at boot time for the frame buffer, although this may be a BIOS configurable option. The loss of system memory has always been one of the complaints some users have had about integrated graphics solutions, but with memory prices being so low today, this shouldn't be a major issue.

The IGP does also support an external AGP 4X slot, so those of you not interested in using the integrated graphics can just drop in another video card which will automatically disable the integrated video. Of course NVIDIA would like you to upgrade to a GeForce3, which does have some added benefit from a driver perspective, as we'll talk about in a bit.

However, this does bring up an interesting thought - would NVIDIA ever intentionally prevent other manufacturers graphics boards from working on their chipset? The chipsets are still too far from completion at this point in time to really find out, but it is something to think about it. With NVIDIA just getting into the chipset market, it probably wouldn't be the wisest move on their part to start off the with AGP compatibility issues. Nevertheless, if NVIDIA begins to take a large potion of the chipset market, it suddenly becomes very feasible for them to go this route in future chipsets.