I know. By 'games', I meant 'user applications'. My intention was to say that the one/two GPU distimction is irrelevant to computer users - they see one card at one price. I invite you to look at 3dprofessor.org, a non-game GPU review site, which says in every review that ATI offers much better value than Nvidia.

OK, ignore the performance per watt. ATI's cards are still the fastest in the majority of applications at most price points. This is agreed upon by most respected hardware review sites. I would say that counts as 'on top'.

As for "fun buzz phrases", where do you think "single fastest GPU" came from? Nvidia PR. It spins ATI's victory into an apparent Nvidia one.

You were the one who brought up one GPU vs two when it makes no difference to the consumer. I'm saying it was a technology decision - ATI could have created a single fast GPU better than the GTX280 but didn't, not because they were inferior but because it was a better business decision.

No, seriously, what features don't ATI have? I can't think of one.

For what purpose?  I know you want to talk about that, but it hasn't been the subject of any of my posts so far.

I don't know why but you seem to want to start an argument with me.

And what just what is the consumer of highend graphics cards typically like?  I would guess that they're the types that would pay the most attention to a single card or crossfire solution.  Of course, I'm sure you have expert studies and strong sources telling you that it makes no difference so feel free to provide them in links please.

I also find it interesting that you say "ATI could have created a single fast GPU better than the GTX280 but didn't."  Would you care to quote sources for this statement too?

CUDA

To first part: I genuinely don't know which part of what I said you are objecting to. Please quote the specific part.

To the second: I am disagreeing with you because I don't like customers falling for marketing and making a suboptimal choice. I'm not trying to defend ATI: I'm trying to argue that they objectively offer the best value because I want people to make the best decision for themselves.

To the last part: I do actually have some evidence. AMD's quad-core CPUs are 'native', i.e. all four cores are on one die. Intel's quad-cores are on two dies of two cores. This is identical to the ATI/Nvidia setup. Yet no one would argue that Intel's chips aren't the fastest. Yes, provably, AMD has the faster "single die" solution. But AMD is clearly not on top because Intel's use of two chips has resulted in better yields and cheaper production.

As for the second thing, ATI could have copy-pasted another 800SPs onto the first design. This would have doubled the die size and the performance, exactly as including two chips rather than one has done. I can't imagine a technical reason why this couldn't have happened.

Fine. There is zero peer-reviewed, expert evidence that Crossfire is no different than monolithic GPUs.

All I can offer are benchmarks that says a HD4870X2 beats a GTX280 at the same price. That's all I ever meant.

http://www.anandtech.com/video/showdoc.aspx?i=3372

http://techreport.com/articles.x/15293

And what just what is the consumer of highend graphics cards typically like?  I would guess that they're the types that would pay the most attention to a single card or crossfire solution.  Of course, I'm sure you have expert studies and strong sources telling you that it makes no difference so feel free to provide them in links please.

I also find it interesting that you say "ATI could have created a single fast GPU better than the GTX280 but didn't."  Would you care to quote sources for this statement too?