Our friends over at Computer Gaming World have come up with a new metric to measure how well 3D accelerators perform with the games they're supposed to boost. This method is way more helpful than scores currently generated through synthetic benchmarks like 3D WinBench. There has been a great deal of controversy about hardware developers optimizing drivers specifically for WinBench so that their numbers would be higher then their competitors - GameGauge is designed to deliver more realistic numbers.
CGW's new 3D GameGauge tests a single accelerator card and then takes the sum frame-rate number of six popular 3D games run on the card. GG comes up with one specific number for the card in question. This is much closer to real-world frame rates than any other test comes close to right now. This should be a benefit for anyone looking for a great gaming card.
During the CGW press conference at CGDC, attended by editor in chief Johnny Wilson, technical editor Dave Salvator, CGW's Loyd Case, and a panel of experts including Microsoft's DirectX guru Kevin Bachus, 3Dfx's Scott Sellers, and id Software programmer Brian Hook, there was a general consensus that GameGauge was definitely a step ahead in accurately measuring accelerator cards.
Wilson said that the major reason for inventing the 3D GameGauge was "to clarify the confusion in the 3D market." He went on to say that readers wanted to know which cards would make their games faster. The CGW mantra with the project was, "What difference will this make for my game?" Because different games use different techniques to enable 3D environments, the idea of having a benchmark that doesn't understand all those new techniques can be frustrating. By using six games the CGW crew thinks show off the true power of current and upcoming 3D cards, the result is closer to what a user can expect to see using a similar setup.
CGW's staff will use the tests on both a lower-end and higher-end system. On the low side is a Pentium 200 with 64MB RAM and a 430TX core logic chipset. The high-end system contains a Pentium II 400 with 128MB RAM and a 440BX core logic chipset. Although the current version of the test only outputs frame-rate numbers to a specific program and adds those numbers together, ZD Benchmark Operations (the Ziff-Davis division that created the original 3D WinBench benchmark suite) will help CGW make the GameGauge more full-featured as it matures into a full benchmark suite.
As games progress toward retail release, CGW is asking game developers to submit games to the CGW team for evaluation and possible inclusion in the GameGauge test. Some new features like multitexturing and triple buffering will become part of the test as DirectX 6.0 releases into the gaming community.
Scott Sellers told other developers to "stop optimizing their drivers for 3D WinBench and start optimizing drivers for games."
And Brian Hook talked about what the crew at id Software thinks of the new test. "We'll take the GameGauge's numbers more seriously than 3D WinBench because many hardware developers come to id and give us astronomically high numbers. With WinBench, you just never know how to take it nowadays."
The numbers from the tests will be published in CGW. And CGW editors will break down the frame rates for each game. If a specific card does particularly well on a game, it will be noted. If any of the cards have any visual problems with a game, the errors will also be reported.
Loyd Case will be using the Gauge in GameSpot's Ultimate Game Machine area to rate 3D accelerators as well.