So, the crap cooler running at 90C is most likely causing throttling of some kind for them to see no scaling when others do.
What these reviews show, 780ti is faster OC. But the ASUS R290 is lame.
Buddy I suggest you read the article and look at his corresponding forum posts in this topic about this test. It was 90C because it required 1.35V for that overclock; Brent Justice also posted that
he tested stability and max overclocks on the card over the course of SEVERAL WEEKS. He couldn't reach full stability at that clockspeed without the additional voltage. What he could probably have done, was a suicide run that lasts 2 minutes and crashes in the 3rd minute. That is quite common for users and reviewers who don't trutly test overclock stability. BTW, It also did not throttle. Had you read the article, you would be aware of this. All due respect, I know this hard for you to accept but the Hawaii just doesn't overclock consistently or all too well. Some websites will get 1075mhz while others get 1175. But Hawaii just isn't consistent despite the fact that voltage isn't locked. The reference GK110 *is* voltage locked without user intervention and overclocks better despite that as far as I can tell. This test seems to collaborate that fact. And the GK110 didn't require 1.35V to get the 1163mhz overclock. But if it had been tested at 1.35V with a techinferno BIOS? The results would have been better for the GK110.
The facts are: it (290X) required 1.35V for that overclock, and this was an Asus press card, and it didn't throttle, keep this in mind. The VRM temperatures were a non factor. At max load the VRMs didn't get too hot as to cause crashes or throttling. In fact, like I said, the card did not throttle. Asus normally isn't in the practice of sending reviewers "dud" cards, either.
So the 780ti wins. This wouldn't be an issue, either, given that the intended price for the 290X is supposed to be lower. Unfortunately it isn't lower for anyone in the states or Canada for that matter.