Introduction

NVIDIA’s grand entrance to chipsets

When the original nForce chipset debuted last year, we regarded it as one of the most significant releases in the history of the system chipset market. While traditional chipset evolution up to that point had followed a slow, conservative, evolutionary approach, nForce brought with it entirely new levels of performance. Rather than rely on the host CPU to power such features that have been found on integrated chipset as audio and networking (and thus drain some of the CPU’s resources), the nForce chipset brought many of these same features without drawing any additional power from the CPU. In fact, not only were these features present on nForce, they brought with them new levels of performance.

On the graphics side, NVIDIA implemented a modified variant of its GeForce2 MX graphics core. Its twin pixel pipeline, 175MHz core offers a fill rate of up to 350Mpixels/second, that’s considerably more power than any graphics processor that had been previously implemented in a chipset. For audio, nForce is no slouch either. Its audio processing unit is capable of generating up to 256 simultaneous voices, and it is still the only audio solution on the market that is capable of real-time Dolby Digital encoding. For connectivity, nForce offers built-in telephony and networking with support for 10/100Base-T Ethernet/Fast Ethernet, and HomePNA 1.0/2.0 phoneline networking.

With so much power inside nForce, it’s no surprise that NVIDIA doesn’t refer to nForce as a system chipset (with traditional North and South Bridges) but as a distributed processing platform. When it was originally launched, each one of the components that made up nForce was more than capable of keeping up with anything that had been offered up to that time.

From paper specs to reality

Unfortunately for NVIDIA, things weren’t as rosy as they appeared on paper. Not only were third-party motherboards shipped later than NVIDIA projected, they were initially only available from two manufacturers: MSI and ASUS. This kept prices artificially higher than expected and hampered availability. Over time these problems were slowly overcome (and NVIDIA released an additional nForce variant without integrated graphics), but our expectations at launch were never fully realized.

So what does NVIDIA do for an encore? Release an updated chipset with support for newer, faster, memory technologies and an entirely new graphics core, and don’t forget to up the bandwidth of the TwinBank memory architecture. NVIDIA has tossed in a few additional goodies and wrapped all this around a new name: nForce2.

nForce2 reference board