Thread: 3dmarks

You can't really rely on 3DMark.

you can't really rely on any benchmark. GPU, CPU or anything

Most companies write specific optimisation code for the common benchmarking tools, and when the device detects it's being bechmarked, switches over to a pre-programmed setting

So much is placed on benchmarks, particularly with graphics cards, that manufacturers will try to duke the stats by any means

Originally Posted by Ynot

Most companies write specific optimisation code for the common benchmarking tools, and when the device detects it's being bechmarked, switches over to a pre-programmed setting

Wow, really? I had no idea.

It's true, Firefox has a 93/100 in the Acid 3 tests, but when you go to a site that uses a lot of different stuff at once, it starts looking nothing like a 100/100 browser.

Lots of times its more about the engine than the graphics card. A well designed graphics engine will always be faster than a poor designed engine running on a better system. Case and point: DOOM3 vs Halo. DOOM3's engine was much better designed and almost never lags, Halo can lag during regular gameplay sometimes, even on a good system.

Most of the time the only way to properly benchmark is to test your system in different high-end games, then compare your average FPS rate to other systems, or simply conclude that running Crysis at max-settings with a stable 60 FPS is probably pretty good.

Good benchmark games are some of the following:
Crysis
Unreal Tournament 3
Left 4 Dead
Portal/Team-Fortress 2/Half-Life2: Episode 2

These games utilize some of the industry's most frequently seen engines, such as Crytek 2, Unreal Engine 3.0, Source Left 4 Dead build and Source Episode 2 build.