6990 or 2 X 6950 ?

[arredondo]

Scientific facts are hard to argue against.

I didn't say no one could perceive the difference. I said the vast majority cannot.

There's always a few oddballs in any large population.

You're right, scientific facts are hard to argue with. And it's a scientific fact that all snowflakes are "pretty unique", so what you technically are claiming is that only one person can see a difference in frame rates above 60 fps. And scientifically speaking, that's completely wrong.

I know a lot of gamers who can see it. I've run tests with my non-computer literate family members and they can see the difference every time as well. In fact, I've never met anyone who has seen comparisons that can't tell the difference.

Maybe you're not a pretty unique snowflake, but there are far less of you than you think.

 

[bunnyfubbles]

Scientific facts are hard to argue against.

I didn't say no one could perceive the difference. I said the vast majority cannot.

There's always a few oddballs in any large population.

Can you point me to the scientific study that used control groups to test if people could see tell the difference between playing on a rig with 120Hz and 60Hz (and of course had the hardware power to push 120fps)? Because I would bet money that a majority would be able to take that Pepsi challenge and get it right.

60Hz was settled upon because it was good enough for the time being, not because it was more than enough (which would actually be around 500-600Hz as a safe overkill cutoff).

For the most part I'd also be willing to bet that people who supposedly can't tell the difference simply haven't ever seen the difference because they've never had the hardware to see for themselves (probably because they never spent the money and/or couldn't/didn't-want-to spend money to buy it)

 

[BrightCandle]

Everyone seems to be focusing on the microstutter question. Its a minor point because its only a problem when the frame rate is already low. I have seen microstutter of 40% (XS has a program for measuring and calculating microstutter) in BF:BC2 in eyefinity 5760x1200. That amount of microstutter nearly halves the frame rate, which at 35 fps gives me an in game effective frame rate of about 20 fps. But its rarely a problem because 2 cards are also considerably quicker than 1 card, you wont see it often and its just another aspect of poor performance you need to account for when determining the settings in a game.

The real issue with crossfire is whether it works in a game or not. If you buy games as they are released you'll find some unplayably slow without tweaks. You'll need to disable crossfire or mess around with the MVPU mode settings in RadeonPro to get it to be playable. Those games that do work and have working profiles often don't scale to 100% extra performance from the second card. Its for this reason I recommend getting a top card first, because you're guaranteed to get that extra performance and sometimes you will be stuck using a single card.

F1 2011 is a prime example of a game that does not support crossfire and which Codemasters say never will. AMD has not yet released a profile (its a month since the game came out) and you wont get much above 18 fps until you either turn off crossfire or set it into D3D AFR mode (which has poor scaling in this game). But its specific to certain games. Keep those drivers and profiles up to date as the profiles come out weekly. If you don't you'll find games that do not run well.

All in all crossfire is hassle. Because I play in eyefinity as much as possible its worth it to me but you should consider whether you want to reinstall profiles every week, use RadeonPro to tweak games into working and measure performance with fraps and calculating microstutter to determine if its acceptably playable. Or you can just get the beefiest single card and have none of those problems.