I currently have two Radeon HD5850s in CrossFire, on the GIGABYTE 990FXA-UD3 motherboard which supports 2-way CrossFire with full x16 bandwidth in both slots. I was hoping to see significant performance gains when I added a second HD5850, but I was quite disappointed with what I got.
In most games like Crysis Warhead and F1 2011, I actually got identical if not LESS frames per second with two cards compared to one. In 3DMark Vantage, though, it did yield a higher score. 15,000 with one card and 22,000 with twin cards. But that is like the only place where I actually saw significant improvements of 5% or more.
Does this have to do with poor Catalyst Application Profiles? I have been keeping up with and downloading all the latest drivers and CAPs, but I still see next to no change in most games. In fact, F1 2011 runs about 10-20FPS worse in CrossFire than it does with a single card, so I downloaded the latest CAP that tweaked this issue, and I might have gotten 5 more FPS than before. Still worse than single card.
Could this be the fault of another piece of hardware though? I feel like the cards are being bottlenecked one way or another. I have a Phenom II 1090T CPU and 4GB of DDR3 2000MHz RAM, with 7200RPM HDD in Windows 7 64. Is there a good, clear way that I can test this rig to see the truth about the performance change due to CrossFire?
Here is a good thread on this, you are not the only person who has this same problem. Does Warhead have a 32bit exe file you can start the game with? This seems to be the largest consensus fix for people with your problem in all the forums that I have read. Running it in 64bit mode seems to throttle the crossfire benefits for some reason, anyway, go have a read. http://www.tomshardware.com/forum/110258-13-crysis-6870...