 |
|
|
How do I get the frame rate in an actual game?Goto our Frame Rate Page to find out how to get the games to report the frame rate. This is an incomplete list, but it is a good start, and it is much more accurate than any benchmark test.
Should we be leery of benchmarks?Yes, be very leery. 3D Winbench will not tell you how a particular card will actually perform in real game situations and does nothing to indicate how a card will handle OpenGL games/applications. The other problem with 3D Winbench 98 is that companies can modify their drivers to run the benchmark test faster without effecting actual gameplay. It is very possible as other tests become more popular (like Final Reality) that these tests may become compromised. The best test is actual real world game frame rates.Tom from Tom's hardware site has a lot of good comments on this subject on his 3D Hype Page. Also, while the GLQuake timedemo is a great indicator as to how a card will perform in Quake engine games, it won't tell you how the card in question will perform in Direct3D games. At what point do the benchmark tests become meaningless in a real world situation?As soon as you play the game. The real test is how does a card performs in your system with the games you like to play. If you look at game play vs benchmark results things just don't add up. This again brings up the Quake timedemo which is a game specific demo. With this test you can not only test the speed of the card, but you can also look at the quality of the image. Different cards produce different looks and feels to the game. Sometimes this is due to a game using a specific API (Glide, PowerSGL, RRedline, etc) and sometimes it is due to the use of different features or larger textures. All of this is important to consider when judging a card.
Do video card manufacturers cheat/tweak their drivers for benchmarks, for individual cards on individual games?If they can they will. ATI, Intel, Nvidia, and VideoLogic have been accused of cheating. ATI seems to be guilty, while Intel and Nvidia are locked in a battle of symantics- each claim the other is not doing anti-aliasing correctly. VideoLogic were optimizing their driver to work better in a particular environment. It was found that the same tweak would work in real world games too. My personal take on this is: whatever works. If you can tweak the driver to run better or faster in a particular game, then that's good to the consumer. The problem comes if the tweak is only to increase benchmark performance. Thus, ATI is guilty and the rest are OK. Benchmark cheating has been around since the Speedstar Pro did it back in '93, and the problem is unlikely to go away. If a manufacturer thinks he can get away with it then they will cheat. Tom, from Tom's hardware site, has a lot of good comments on this subject on his 3D Hype Page. Does 3D gaming have benchmark results?Not any more. Frankly so many other sites have benchmark results that it is useless trying to keep up.
If you really want some benchmarks try Http://www.madonion.com. One last note: If you want to see which card is faster then, play a game with it!
|
