If we're putting in guesses I'd guess Polaris 10 will be 20-50% better than 390X based on the supposed 130W TDP
I hope this is true but from what most have said, 150 watts is going to be the upper end based on the 6-pin configuration, it remains to be seen how close AMD would want to come to that limit.
Whats the closest we've ever come to the 6-pin power limit. Has AMD/NV ever released a 6-pin card with a TDP of 150w? Really curious about that one.
There's also the Primitive Discard Accelerator to consider. And improved shader efficiency, which may be power efficiency or hardware usage. If Pascal's frequency improvements are a result of the process, I expect Polaris to have similar, or slightly worse, frequency improvements but probably higher relative base clocks than GP104 prodducts.
I think based on what I've read about the processes we can expect GloFo/Samsung to have a ~10% density advantage.
Apple spares no expenses designing its chips, after all; they are some of the highest grossing chips in human history! I have no doubt Apple had two dedicated teams for the two fabs and their respective designs for A9. Here we see Samsung with a 9.2% density advantage.
Copied from Chipworks
TSMC 16FF vs Samsung 14LPE
From Tomshardware
http://www.tomshardware.com/news/iphone-6s-a9-samsung-vs-tsmc,30306.html
Based on the results of our testing, it's clear that both versions of Apple's A9 SoC deliver the same level of performance, but Samsung's 14nm FinFET process appears to offer slightly better power efficiency, extending battery life between 3.5-10.8 percent.
So according to them, Samsung's process has better leakage, which would make me think they can clock at least as well as TSMC. However, this can't be assumed too much as Apple most likely designed for lower clockspeeds as they have always done.
Remember, this is all on 14LPE. We know AMD is using the Gen-II version of Samsung's FinFet 14LPP which should improve clockspeed/leakage even further.