The more ridiculous the more consumers will buy would be the illogical argument? Like the worse your product the more people will buy it.

I take no personnal offense, just want the justification of why you think that or what was your intent... since PS4 was the strongest of the 3, wouldn't be the buyers of the other systems that would buy even a two hamster system if by they company of choice?

Sadly there will be no elaboration on this, or any other aspect on previous posts I have made on this thread. If you took personal offense to this or any other post I have made, I apologize.

It's most certainly possible. Global Foundries and AMD would have learned a bit from Polaris (14nm is still new, power characteristics are still up in the air for the most part.) one would expect those lessons would translate towards improving AMD's Semi-Custom chips and Vega/Zen.

Would this be a likely outcome though? I can't really say without more information. But there are two ways to cut the tree down, throw more hardware at the problem... Or higher clocks.
nVidia this round went with higher clocks rather than more hardware this time around as the Geforce 980Ti is an 8~ Billion transister chip, the Geforce 1080 is only 7.2 Billion, but has a 60-70% higher clock.

but they did that going from 28nm to 16nmFF - doing something similar while using the same fabrication node seems rather unlikely to me

it might be a possibility if they go to TSMCs new 10nmFF node, but I'm pretty certain that one is quite a bit more expensive than GloFos 14nmFF

Not sure what part of my point you are arguing against, nVidia will likely release a larger lower clocked chip in the future at 14nm (We are stuck at this node for a LONG time.)
The 1080 is built on a smaller process, so it should be able to have more transisters than something on an older process. - Well that's the trend we have seen over the past several decades. New process, smaller chips with more transisters.
Thus far we have new process, smaller chips, less transisters.

The Xbox One and Playstation 4 are currently built at 28nm. The Neo and Scorpio will be 14nm, but even at each node you still have wiggle room in terms of voltages/clocks/transister counts, hence my entire point.

Heck. Let's say Scorpio has the exact same GPU as the Neo, Microsoft will need a clock rate of 1300Mhz to hit it's theoretical floating point performance, Polaris easily manages that when it has the appropriate power delivery feeding it.

We know that Global Foundries is using Samsung technology for it's 14nm, complete with samsung tooling.
Which also means that they have LPP and LPE, for all we know Global Foundries use LPE/Low-Power Early for Polaris... Where-as LPP or Low-Power Plus also allows for an increase of 15% in performance whilst also saving an extra 15% power which the consoles might be using.

Anyway. My entire point is... Other than a couple of arbitrary "Teraflop" numbers and a few rumors which may hold zero basis in reality... We have zero idea untill the official unnouncement, if you wish to argue against that... Be my guest. :P

Perma you seem to be not understanding or trying not to. We are talking at Scorpio and Neo using the basic same architeture, and you even want what they learn producing Neo to make Scorpio cheaper (and in some bizarre way even cheaper than Neo). So I'm not talking about making the chip smaller with a node reduction. We are talking about both systems using same architeture on similar node process and you wanting one to be smaller, cheaper, more powerfull, etc. You still haven't explained that.

Because we certainly know the node reduction have made PS3 and X360 cheaper.

Are you suggesting that lessons learned by AMD and Global Froundries with getting better power characteristics and yields at the Fab level will not benefit everyone?
Because that would be false. And I assume that is what you are arguing against.

And I get we are talking about chips using the same architecture on the same process. I have explained how it would be possible for one to be smaller, cheaper and more powerful.

I'm suggesting that not ALL leasons learned will carry from one to the other because of the specific process... PS4 and X1 use the same architeture and still PS4 is cheaper to produce because it have a bigger bulk and got more cost cut due to it and leasons learned because of it, while X1 got some of the leasons learned from PS4 it didn't get all because both chips have differences.

How are you saying about the same process and architeture while assuming PS4Neo gets the bigger node and Scorpio get the smaller node, because of reasons?

I never said that Scorpio get's a smaller node than Neo. - Perhaps you are getting the terminology confused?

And you are right. Most non-fab related lessons/improvements will not carry over between the two, but that's also not the argument we are having.

Also the Xbox One chip as far as I know is larger than the PS4's, so it's more expensive to manufacture on size alone.
However... Not sure if the eSRAM can get away with cheaper patterning though as it's an inherintely simpler and predictable structure just like NAND.

So if both are on the same node, how will it get cheaper and faster transistors for it to be smaller, needing less space and CUs?

My arguments is that since both cards aren't exactly equal not all leasons will be transferred, can be 70 or 90% but not 100% because some improvements will be dependent or on the different aspects of the cards and if Sony order 3x more units than MS then they will get more bulky reductions .

Yes I agree Xbone is more expensive because of Size and eSRAM help on that cost (not only because it makes the chip bigger, but also is another part added).

But Sony chip isn't smaller due to having less but more powerfull CUs, and probably because having better transitors to reduce the size of the card would mean bigger and not smaller costs.