Problems with ATI 5770 HD Crossfire Performance

Hi,
I am having some issues with my PC and would really appreciate any insight!
First off, my setup:

CPU: AMD Phenom II X2 555, 3rd core unlocked; CPU-Z says it runs at 3.2Ghz under full load.
Memory: 16Gb DDR3
Video: Two Sapphire Radeon HD 5770 1GB 128-bit GDDR5 cards in Crossfire; the 2nd card was added maybe a 18 months or so after getting the first one; this goes to my receiver and my TV running at 1920x1080
Mobo: ASUS M4A87TD EVO AM3 with 2 2.0 PCI-e slots (16x, 4x)
HD: My main one, where my Steam files are, is a WD Caviar Black 1TB 7200rpm
OS: Windows 7
Catalyst Control Center (CCC): v13.4; hardware info shows both the primary and linked adapter (primary running at PCI-e 2.0 at 16x, the linked card running at 4x)

I just updated from CCC 12.4 (or 12.8?) and I have had Crossfire enabled every since installing the second card. With CCC 12.4, and every previous version of CCC I've had, about 60% of the time when I booted up from shutdown, I'd have to go back into CCC and re-enable Crossfire; for some reason it would be disabled on restart more often than not. I've only had 13.4 for a day now and I have not yet shut down my PC.
My main issue is basically I have never really noticed any difference from running in Crossfire vs running one card when gaming (Crysis 2, BF3, Portal 2, others). I think I am fooling myself into seeing a difference, but sometimes I would start up my system, start gaming (and it seemed normal), then remember that I didn't check if Crossfire had been disabled so I would recheck and see it was disabled and so I would renable it, and then go back to gaming and not really notice a difference. Does that make sense? And sometimes it would even seem worse.
So I bought Hitman Absolution on Steam (not pirated) and noticed that it had benchmarking which I've never done. These are my in-game settings:

Resolution: 1920x1080
Fullscreen checked
Refresh Rate 60
VSync checked
MSAA 2x
Exclusive Fullscreen checked
Aspect Auto
Quality: Custom
Shadow Medium
Texture Quality Normal
Texture Filter Anisotropic 2x
Reflections Medium
FXAA checked
Global Illumination unchecked
Tessellation unchecked
Level f Detail Medium
Depth of Field Medium
SSAO Normal
Bloom Normal

With my previous version of CCC my FPS averaged around 24.8, but I didn't write the exact numbers down. I went to the AMD website and saw a new CCC version and the new CAP with Hitman Absolution. I uninstalled all previous CAPs, and ran the 13.4 installer to remove my previous version of CCC. Then I rebooted, made sure display drivers were all uninstalled, and installed 13.4. I ran benchmarks with Crossfire on and off, then installed the new CAP and ran benchmarks with Crossfire on and off (each three times, "A", "B", and "C"). Here's my avg FPS:

Crossfire Enabled, no CAP
A. 26.3fps
B. 28.13fps
C. 27.8fps
Crossfire Disabled, no CAP
1. 29.9fps
2. 30.5fps
3. 30.5fps
Crossfire Disabled, CAP installed
1. 30.3fps
2. 30.3fps
3. 30.3fps
Crossfire Enabled, CAP installed
1. 16.4fps
2. 16.3fps
3. 10.4fps!

So it almost looks like my best settings were with Crossfire disabled and no CAP. So....Why? I mean, if I enable Crossfire and I do what AMD suggests and use the latest CCC version with the lates CAP, I get 10-16fps! I am not too technically inclined, so I was thinking I either have a faulty 2nd card (although CCC says it's linked) or my 4x PCI-e slot is slowing everything down, or I really don't know.
If you have any ideas/suggestions please let me know!
 
Hi,
Thanks for your reply! Tried your advice with pretty much same results.
CAP = Catalyst Application Profiles. These are .exe files made by AMD that are supposed to be in addition to CCC that add some profiles or something to CCC that optimize the use of Crossfire. They release CAPs periodically after a few games come out. For instance, the latest CAP has Hitman Absolution support, and this is supposed to optimize the drivers to run with Hitman and several other games (or something along those lines).
CAPs are located at http://support.amd.com/us/gpudownload/windows/Pages/radeonaiw_vista64.aspx
Thanks again!
 
To op.
I think you are getting better performance with a single card,because the cards are both working @4x when in cf.For cf to work I think both cards should be getting an equal data feed .So 4x will not cut.
Not entirely sure though.
 
To op.
I think you are getting better performance with a single card,because the cards are both working @4x when in cf.For cf to work I think both cards should be getting an equal data feed .So 4x will not cut.
Not entirely sure though.
Yes, check out the specs here: http://www.asus.com/Motherboards/M4A87TD_EVO/#specifications.
It says your board has "2 x PCIe 2.0 x16 (*blue@ x16 mode, black@ x4 mode)", so you have 1 x16 really... bad wording. Thanks for the expansion pistopito. I think the PCIe bandwidth is your problem. But did this occur with 12.4/12.8?
 
Back