I agree on the premature ... was essentially opening line in my original post
I also agree that nVidia's Boost 3 has put the kabash on overclocking in the 10xx series, making it difficult for AIB partners to distinguish themselves from one other. Reviewers have been consistent, and the temps support this, that the 10xx series cards have been intentionally "handcuffed" and can do much more. And for the 1st time, nVidia has tightly limited access to the BIOS and no "crack" has yet appeared which would allow them to be manually edited.
This happened last gen when the 970's performance was intentionally nerfed by lowering the throttling point of the 970. So what's the deal with 10xx ? ... why has nVidia intentionally gimped their cards with Boost 3 ? Given their recent history, their modus operandi has been to keep something "on the shelf" so they can react to what AMD does .... When the performance of the AMD cards became known, it was put forth based upon changed specs that the "planned" 770 became the 780 and the 780 was "shelved" because a) the lesser card was cheaper to make, it outperformed the other team's flagship and c) by putting it on the shelf, they had a "ringer" ready to go and sitting "on the bench". AMDs entire ad campaign for the new card line was hurt badly .... as as soon as the announcement was made, days later Vidia stole the spotlight with the 780 Ti and all the "new stuff from AMD" chatter died.
But now ... since nVidia has two cards above the 1060 / 480 out and AMD doesn't... what can they "spring on the market" quickly when the AMD cards do come out with the advantage of a longer development and production cycle ?
For 3 generations now, AMD has been very aggressively clocking their cards in the box. Where nVidia has "stayed the course" offering 18% to 32% OC capability .... thru 10xx that is. Meanwhile AMD for the most part, remains stuck in single digits. Of course there is a risk of aggressively overclocking cards in the box ... more cards that just won't hit that number which means, for those that notice ... more RMAs (like my son's EVGA FTW). So by gimping the current cards with Boost 3 to keep them just above the competition, means less RMAs ... having a "switch" that they could flip if and when new cards from AMD arrive, might mean an increase in RMA rate comparable to AMD, but it would help them maintain performance rankings. Kinda like a car race where in a close race driver ups the rpms to maintain the lead ... but of they far back in the rear view mirror, no need to risk that.
One of the simplest things they could do if they find themselves behind is to "take the handcuffs off" and let the AIB cards shine. More on this later....
overclocking pascal has been very disappointing and not a single one of the "super" cards has been able to top what the cheaper ones can do
I have to disagree here ...
1. The supercards final death throes came with the 9xx series....but the cards diagnosis was terminal with 7xx. The 7xx served to portend what we know today; the Classy, Lightning, Matrix 7xx series were not delivering significant performance increases which could justify the increased cost ... yes you could get higher clocks stable (custom BIOSs helped here), but they weren't delivering higher fps even with the higher core speeds. The 1-2% was just too paltry to justify the $100 price increase.
The value originally with these cards was what **you** could do with them, not how they came out of the box ... nVidias legal and design limitations are still growing increasingly tighter but drastic leaps already arrived with 7xx and 9xx. Boost 3's approach is different in that it gimps everything and all the well designed AIB PCBs are looooong ways away from throttling. What if "all of a sudden", those BIOS entries were changed to allow more voltage and higher temps ? With 10-15C, they have lotat room to work with.
2. But the super cards are not what I was referring to ... we're talking "overclocking" as opposed to "buying factory overclocked cards". When I say overclocking percentage, I am not talking about the fact that the MSI 1070 gaming is 4% faster than the reference card. I am saying that the MSI Gaming when overclocked manually is 17.7% faster than the reference card when is when the reference card is taken outta the box. The problem with overclocking is you have to guarantee the results
Yes, all of the factory overclocked cards are close to one another but
I don't see how 18% more fps is not a "good overclocker" ...especially when looking at AMD's results for which anything in double digits is a great rarity. But also remember that the 980 Ti was 30% when the 970 was in the high teens - low 20s. It's the flagship card and as ssuch, they loosened the OC reigns quite a bit. Yes you hadda pay for that big OC headroom and the more ya paid, the more ya got... 970 just under 20%, the 980 was 25% and the 980 Ti was 31+%.
I can't help but suspect, that ..
a) the 1080 Ti, originally rumored for Q4, 2016 won't drop until AMD releases their answer to the the 1070 / 1080.
b) When it does happen... and if AMD is able to make inroads against the 1060 / 1070 / 1080, I think we will see "the handcuffs come off Boost 3" and we'll no longer see 10-15C of headroom between GPU temps and the throttling point on the AIB cards.