All of those supposed GTX 690 clockspeeds you mention are impossible because they are done in specific 13mhz increments. 1200 is not a possible clockspeed. Neither is 1080. You're estimating, though, right? Any valid explanation for clockspeeds that aren't possible on the Kepler?
Here's a fun fact: what's the advertised boost speed of the GTX 690? I'll let you figure this out and get back to me. (here's a hint: far lower than any clockspeed you mention) The 290X struggles to even hit it's boost at factory defaults, the Kepler GPUs far exceed their advertised boost at factory defaults. You're backing up that point with the GTX 690 even if you're throwing out clockspeeds that are not possible with the Kepler.
Well I'm sorry I didn't write down the exact clock speeds when I had the cards. Go ahead and be pedantic I rounded the numbers off. Clock speeds is not the issue here. It's how much variation the is with different cards. you are claiming kepler cards have a variation of 1%. Thats just not true and I don't know how you believe that.
So if the boost clock is 1000mhz and the review cards are boosting to 1.5ghz, but retail cards care boosting the 1100mhz. Thats fine in your book cause they are boosting higher than the numbers on the box even though the card is a lot slower than the review cards?
But if the AMD card has a boost clock of 1000mhz and the review cards run a 900mhz and the retail cards run at 850mhz. That not ok, because they are below the max boost clock?
Is that the problem here?
Not that review cards are faster than retail ones, which is possible with both hawaii and kepler? but because the numbers on the screen don't match the ones on the box?
I still don't see this 15% performance drop. You keep bringing graphs that show the 290 droping 300mhz, but I don't see any actual gameplay performance detriment because it looks to me like all the reviewers are warming up the cards.
While you were fussing over the clock speeds i listed. You missed my point where i stated that Kepler has JUST as much variation as hawaii and there are reviews that back this up. That variation being about 5%. But you keep pulling this 15% from somewhere. If you are going to say it is from 1 run to the next. Kepler behaves exactly the same way and you know it. It is mentioned everywhere. It's starts at max boost and drops as the cards heats up. As much as 20% as shown. GK110 is just as bad. As it's seen by adding fans blowing cool air over the cards alow it to sustain boost longer and higher. Same as Hawaii.
I urge you to read this review. The Asus card should be beating the MSI and Gigabyte cards, but it isn't. Why is that? because of variation. That is not 1%.
There are plenty of other European sites that have done this type on testing on kepler. I don't know if you've just missed them all or avoided them.
The only thing you can blame AMD for here is the terrible cooler.
EDIT: Wait a minute, I don't even know what exactly the problem. Do you have a problem with card to card variance or do you have a problem with the cards not running at 1ghz all the time or do you have a problem with press cards being faster than retail cards?