Crossfire and FurMark

[wolfram23]

Hey can anyone answer me this. Why does crossfire barely affect FurMark results? I tried 0xAA and 4xAA with only about 2-4 avg fps increase (multi gpu version btw). Now, in gaming the advantage is very noticeable. I'm running Crysis at Very High 50fps no slow down (2xAA - works just fine at 4xAA but with slow down using binoculars but 2xAA is buttery smooth and I can't tell the difference). Stalker Call of Pripyat as well, I boosted all the settings to their max and it's running great. It's just in GPU stress tests the results are pretty much negligible. I'm not going to let it bother me but can anyone give some indication why this is the case?

 

[shadow187]

ATI automatically throttles their cards in Furmark, because they think that program is a 'Power Virus'. Don't worry about it.

 

[Rustyy117]

maybe your running Furmark at a low resolution and your CPU is bottlenecking the performance. Just an example of what I noticed in RE5 (Resident evil 5)

Core i7 920 @ Stock 5850@ 1000/1200 All high 8xAA 1280 x 768

Average 103.6FPS

Core i7 920 @ 4Ghz 5850@ 1000/1200 All high 8xAA 1280 x 768

Average 141.7FPS

Thats a 26.9% increase. With your two cards the bottleneck will start at higher resolutions. What I suggest you do is set Furmark to your highest resolution and test with one card, then test with crossfire and compare the results.

 

[rofl_my_waffle]

crossfire works better at higher resolution. The extra card may be worth 30% if you are using small resolutions like 1280x1024 or as much as 60% if you are using 2560x1600.

 

[wolfram23]

Good point! I was running 1280x1024 windowed so I could watch the ATI graphs real-time. I'll do it over at 1920x1080.

 

[wolfram23]

Best answer selected by Wolfram23.