Ignoring The Speed Limit
Nowadays, buying video cards is no longer as spectacular as it used to be - aside from the very steep price. The cards themselves differ mostly visually, in that they may have a colored PCB or heatsink, or maybe an extra large fan. Other than that, boards that use the same chip are largely identical. What really sets them apart from each other is usually their feature set. Where one card offers a digital output for flat-panel monitors, another comes with TV-In and/or TV-Out, 3D shutter glasses or hardware monitoring. Lastly, the bundled software may be more of a deciding factor than the brand name.
From a performance perspective, all cards using the same chip are on the same level. A no-name GeForce2 Ti will be just as fast as a Gainward GeForce2 Ti 500 with extra fast memory chips, for example. This phenomenon is easily explained. The deciding factor in a card's stock speed is the card BIOS, which sets the memory and core clockspeeds. Until now, NVIDIA has always kept a very close eye on the card makers, to ensure that no card was set to run outside the specifications set forth by NVIDIA. Gainward is attempting to circumvent this problem by overclocking through the driver - which is no different than overclocking through software tools. ASUS has introduced a similar "feature" with the "3D Turbo" options. ASUS' solution is nontransparent, though, in that you have no real control over the level of overclocking. But these cards will only run at higher speeds with the company's own driver - switching to a reference driver will naturally reset all frequency to the standard as set in the card's BIOS. Only with the introduction of ATi's RADEON 8500 has NVIDIA's stance on overclocking changed. Suddenly, NVIDIA's official company line is that no one is really opposed to higher frequencies and a little tweaking. Competition: It's what's having you for dinner!
Still, as things stand now, features like faster memory chips or extra cooling only come into play when the card is overclocked manually, be it through software or an "unofficial" BIOS. As our overclocking tests show, the results are well worth the effort, as most cards still have a lot of headroom. Unfortunately, most companies still officially oppose overclocking, fearing consumers could damage their cards and try to get replacements through warranty programs - a very costly prospect. Only Gainward openly favors and even supports overclocking, guaranteeing stable operation when a card is overclocked - within certain limits, of course.
Looking at the current generation of video cards, it is obvious that choosing the "right" card has become more a matter of taste than of technical finesse. All companies base their drivers on NVIDIA's reference release. Most don't even bother to replace NVIDIA's logo with their own anymore. The American companies seem particularly unimaginative in this respect. Most cards from Taiwan at least come bundled with a small collection of tools or utilities that make everyday use of the card more comfortable or allow certain tweaks.
With the cards being as similar as they are, the only other remaining point of interest (besides overclocking capabilities) is a card's signal quality. For all cards in this group, the image quality ranged somewhere between good and very good - depending, to some extent, on the monitor used.