2900 XT VS 8800 Ultra question

[decoy26517]

Ok, don't flame me for this, its just a question:

I was going some research on these two cards; the ATI 2900XT and the 8800Ultra, and I check futuremarks website for comparisons as well as Toms benchmarks for comparisons and I was wondering why there was such a huge difference.

On Futuremarks ORB (Online ResultBrowser) the 2900XT compares really well to the 8800Ultra, and sometimes even beats it in crossfire configuration. but on other benchies I find that the 8800 actually does alot better then the 2900. Why is this?

 

[justinmcg67]

because 3DMark06 is not a very accurate representation of real-world performance. Look at ANY benchmark, an 8800GTX Ultra will beat the living tar out of an HD2900XT. But in 3DMark06...it wins. I've never used 3DMark06 as a scale of actual performance, just a scale of how developed a GPU is. The HD2900XT has a ton of stuff packed into it, but, in applications, the Ultra dominates.

 

[decoy26517]

But thats the thing, even in Toms benchies when their using 3Dmark06, the 8800ultra beats the 2900XT by alot. Except in systems listed on 3Dmarks website (which are more real world IMO then the test machines during benchies) the opposite is happening.

 

[lunyone]

Consider the 3dmark series test like the EPA rating on a car. You may reach this or that result, but in real games you'll have different results. So 3dmark series is a synthetic benchmark and not a real world game benchmark.

 

[Heyyou27]

Ok, don't flame me for this, its just a question:

I was going some research on these two cards; the ATI 2900XT and the 8800Ultra, and I check futuremarks website for comparisons as well as Toms benchmarks for comparisons and I was wondering why there was such a huge difference.

On Futuremarks ORB (Online ResultBrowser) the 2900XT compares really well to the 8800Ultra, and sometimes even beats it in crossfire configuration. but on other benchies I find that the 8800 actually does alot better then the 2900. Why is this?

It's called a synthetic benchmark for a reason; real world games will stress different parts of the GPU more than benchmarks which basically test only certain features. If you were to increase the resolution to 2560x1600 and enable 16xAA and 16xAF on both cards, I'd bet the HD 2900XT wouldn't be nearly as close.

 

[toasty2]

Also, I believe Tom's Charts are out of date since the latest drivers.

 

[decoy26517]

Also, I believe Tom's Charts are out of date since the latest drivers.

Thats what I'm thinking.
I wasn't trying to compare game/other performance with synthetic but compare synthetic vs synthetic and something wasn't adding up. I guess a lot of people might of misunderstood me.