Widescreen Gaming Forum

Hi guys, I feel like I think this is the best place to post this as I feel like I am digging myself deeper and deeper into an obscure niche by going with a cross-finity setup, and normal solutions just do not seem to apply to me.

I have two 6950's, an Asus EAH6950 2GB, and a reference 2GB 6950 from VisionTek. The VisionTek is my old tried and true card, the Asus is the newcomer. I bought the Asus because I wanted the 4 DisplayPorts, as it seems to be the only remedy right now for the infamous screen tearing due to mixed output ports on stock 6950s. For those interested, the Asus does fix this if you have monitors with DisplayPorts. I no longer have that annoying tear-line on the 3rd monitor.

My monitors are setup in 3x1 Portrait, with a 4th extended monitor in landscape. My resolution is 3240x1920 on my primary display group. Here are my system's relevant specs:

i5-750@stock, w/8GB
Asus P7P55D Evo, PCI-E 8x/8x when crossfired
Corsair 850TX
2x 6950 2GB w/11.8 Catalyst + 11.8 CAP3
Win7 x64

I have both cards installed, crossfire enabled, and both GPU Z and RadeonPro shows activity on both cards during games. This is where it gets interesting. In 3Dmark11 Basic, crossfire works as expected. My results with one card is roughly 24 FPS during the first test, which scaled up to roughly 45 FPS with 2 cards. This is the basic version of 3DMark11 so the test runs at 12x7 resolution. Test 2 shows similar gains, going from 25 FPS average to 48 FPS average, single and dual cards respectively.

However, in actual gameplay, my results are different. In Borderlands, I saw a drastic framerate drop in crossfire, which is as expected, because a quick google for borderlands crossfire returns a lot of hits about it not working. However, when I played Hard Reset Demo, my FPS is roughly halved with crossfire on. On Ultra settings with MLAA@3240x1920, one card gets me roughly 50 FPS. With the same settings, I get roughly 23 FPS with 2 cards in crossfire. These numbers are read from the radeonpro fps counter and taken as soon as the game loads, from the begining. I do have "Reduce Input Lag" disabled in options.

Also, I notice that crossfire appears to be working during the main menu of Hard Reset, the one with the robotic arms lifting panels for options. In that screen, I get roughly 110 fps with crossfire and around 70 with a single card. It isn't until I load the game and get into the actual level that I see the framerate get halved.

I figure this may be a problem with profiles, but I see similar results in Hydrophobia, which I believe did get a crossfire profile some time ago.

So, is my crossfire working? Not? Working with exceptions? Does anyone else with a similar setup care to try out Hard Reset Demo and report their results? All you have to do is to start a new game, let it load and read the fps numbers you get immediately upon loading into the level, with and without crossfire.

How much your cards loaded? Do you have MSIAF to see gpu usage?