3dmark not good?

[Scali]

First go look up logic in the dictionary.

Look up 'fallacy' in the dictionary.

Now if you want to know how well your card will play games then LOGICALLY you will use games to test your card. And (as you pointed out) every game IS different that's why no synthetic benchmark will be very accurate in judging true game performance (especially with many games using OpenGL instead of DirectX)

See the fallacy right here. Every game is different. So how can you speak of 'true game performance' in the first place? Which game is the 'true game'?

Another fallacy there is that OpenGL vs DirectX would make a large difference. They're both running on the same computer system with the same 3d acceleration hardware. Well-written code will not be limited by API/drivers, but by the GPU or CPU.
The differences between OpenGL and DirectX are often far smaller than the differences in game engine design, so they are basically irrelevant.

However changes in your 3dmark score may not actually mean changes in real world game performance.

Luckily there is more to 3DMark than just the overall score. You can get statistics on the various benchmarks, which allow you to draw some conclusions about what the overclocking affected exactly. From there you can extrapolate that real world game performance will change accordingly in similar situations (which DO exist).

 

[PRIME1]

See the fallacy right here. Every game is different. So how can you speak of 'true game performance' in the first place? Which game is the 'true game'?

Another fallacy there is that OpenGL vs DirectX would make a large difference. They're both running on the same computer system with the same 3d acceleration hardware. Well-written code will not be limited by API/drivers, but by the GPU or CPU.
The differences between OpenGL and DirectX are often far smaller than the differences in game engine design, so they are basically irrelevant.
.

Again more misinformation.

A GPU is nothing without it's drivers. A card may perform well in DirectX games but fall behind in OpenGL games all because of it's drivers. In fact a card may perform well in 3Dmark but fall behind in other DirectX games because of it's drivers.

Every game IS different that's why you need to look at the benchmarks for the games you play. If you "play" 3dmark then there you go.

I'm not knocking people for using 3dmark. It's up to you want you want to use. I however find it irrelevant with so many true games that can be used as benchmarks instead.

To each his own.

 

[Scali]

Again more misinformation.

A GPU is nothing without it's drivers. A card may perform well in DirectX games but fall behind in OpenGL games all because of it's drivers. In fact a card may perform well in 3Dmark but fall behind in other DirectX games because of it's drivers.

Talking about misinformation. The era of bad OpenGL performance is long behind us, since OpenGL's API is now so outdated, that nearly everything is done through extensions, which are often based on Direct3D's functionality.
Also, after NVIDIA's batch-batch-batch presentations, most developers are smart enough to avoid driver bottlenecks altogether.
It's not 1998 anymore.

Every game IS different that's why you need to look at the benchmarks for the games you play. If you "play" 3dmark then there you go.

You are not acknowledging the wealth of info that 3DMark can provide, and that pisses me off.
3DMark is an excellent tool, some people just don't know how to use it.

 

[trinibwoy]

Geez, horribly old news.
I suppose nobody realizes that S3 supports this feature aswell? And other vendors will probably support it aswell in their next generation?

Yes, anybody who frequents B3D or keeps up with this stuff knows this is old news. But the article is "new" as in just posted. It was relevant to the topic so I decided to post it - geez you online types are finnicky aren't you.