Related Articles

Background

Visiontek Geforce 2 Ultra

Last April, we saw the release of NVIDIA’s Geforce2 GTS. Produced using the 0.18-micron fabrication process, the Geforce2 GTS had a core speed of 200MHz and a memory speed of 166MHz which was up from the 120MHz /150MHz (core/memory) speeds of the original Geforce (which used the 0.22-micron process). The smaller die process reduced power consumption, which reduced the amount of heat it produced, allowing it to reach the higher clock speed. This also alleviated the power consumption problems that the original Geforce had. The GTS’s four rendering pipelines could process two textures per pixel, per pipeline giving it a theoretical fill rate of 1.6 gigatexels/s (200MHz x 4 pipelines x 2 textures = 1.6gigatexels/s). Compare that to the original Geforce’s 480megatextels/s and you realize how much potential power the Geforce2 GTS actually has. I say potential because the Geforce2 GTS’s peak memory bandwidth of 5.3GB/s limited the amount of processing that the chip could actually perform at any given time. Despite this, the Geforce2 GTS was quite a powerhouse, but it still had plenty of competition to vie for the performance crown.

Not content to rest on their laurels, NVIDIA has upped the ante with their Geforce2 Ultra, which is a large increase in terms of both performance and price. The Ultra chips use the same architecture as the GTS chips, but are handpicked to run at 250MHz /230MHz (core/memory). These speed increases bring the texel fill-rate up to 2 gigatexels/s, and the memory bandwidth to a magnificent 7.36Gb/s (excellent!). This added memory bandwidth gives the Ultra an edge over the Geforce2 GTS in more memory intensive operations (such as high resolution 32bit colour and FSAA). NVIDIA has successfully blown away the competition with their Geforce2 Ultra, putting the performance crown into their hands.

Before we dig into the Ultra any further, I will briefly mention the two other versions of the Geforce2: the MX and the Pro. The Geforce2 MX is the “value” line that boasts fairly decent performance at a much lower price than its higher-end siblings. The MX is the only design that is architecturally different than the others as it has only two rendering pipelines instead of four. The other major difference is that it has built-in support for TwinView – NVIDIA’s answer to Matrox’s Dualhead technology. On the other end of the spectrum, the Geforce2 Pro splices the gap between the GTS and Ultra models, with performance above the GTS, but below the Ultra. It shares the same clock speed with the Geforce GTS, but its memory runs at 33MHz faster which solves the memory bandwidth issues. If you haven’t heard about the Geforce2 Pro, it is probably due to the fact that it was originally destined to be an OEM only solution, so its launch was very quiet. Together, these four versions of the Geforce2 cover all the bases of the video card market.

send article   hardware newsletter   article comments (2)

Submit our article to: Digg this! del.icio.us