We have all the clarifications we need.

So far, reviews show that all the AMD RX480 cards draw around 160W.

This settles the question how fast the Sony SoC will be. Even the rumoured/leaked 911MHz gpu clock probably is at the high end of what goes.

I'm hoping the Neo is closer to 6tf just for the fun of watching MS suddenly make a random Tweet about how 6tf was a 'conservative' estimate and it could be closer to 8tf.

Looking at the power draw of the newer AMD cards (just slightly better than Maxwell on 28nm) there is about zero chance of that. MS is probably also in for a nasty surprise.

But both could claim the consoles are close to, or exceeding 6 TFLOPS while the GPUs very very rarely come close to max clock speeds, for marketing purposes. Just like the R9 Nano is clocked the same as the Fury X with much lower TDP.

The gap between Scorpio and Neo will probably be relatively close (think R9 290 vs. R9 Nano) and mostly defined by TDP.

AMD has stated that the power draw of the Radeon RX 480 is not as it was designed and will be rolling out updates to rectify it.

Also keep in mind that comparing a discreet GPU to a SoC/APU is like apples to oranges, ESPECIALLY in regards to power consumption.

Er. The reason why the Nano managed to get a much lower TDP was because AMD took the fury Chip... And tested them all, all the chips which hit the lowest voltages was reserved for the Nano cards. The chips which didn't, were sold as Fury.
Voltage has a direct relationship to power consumption in processors.

Consoles don't have that luxury, they cannot take a chip and scale it's clocks and/or voltages and sell them off to different market segments, they need to be conservative and try to get as many usable chips at a certain clock as they can to keep costs low (Consoles are cost sensative remember!)
Which means voltages are likely to be conservative and thus relatively high, which impacts TDP.

They can get around that somewhat due to consolidation... There aren't multiple memory controllers, power delivery systems, memory pools, you name it.

I'm stepping out for a while. There seems to be people discussing this that know a hell of a lot more than me and that makes me feel inferior. I must protect my insecurity.

A quick question: I know nothing about tech but occasionally use a tech word in the hope of looking like I do, so ... Am I right in assuming that a closed system (console) can get more benefit from a chipset than your average PC? I'm beginning to wonder - because of MS announcing 6tf and their push for PC/XB unity - whether the Scorpio could in fact be a PC and not a console. I would very much appreciate a response from the more knowledgable posters above.

Thanks.

What is it with people these days? Now its power supply? smh

A discrete GPU recieving power is recieving power for eveerything in the GPU. Chip, memory and fan and any other electrical component in there. I mean memory alone can draw up to 20W!!!! And are you aware that even if the PS4 uses around 90-150W in total poeer right now its PSU is rated at 230-250W?

There are a lot more to these things that the superfluous stabdards most here seem to just go with. 

Simple answer, is absolutely. And for a number of reasons. Though things like direct x, mantle...etc on the PC end aim to get more out of the chips but you will still get more oit of any console chipset than with a similar chipset in a PC. There are a number of inherited redundancies on the PC side, basically its strenght of being able to hash nearly any component together and get them to work is also its greatest weakness. Lots of legacy  stuff have to be carried over to ensure compatibility. 

lets not also forget that on the PC end, under it all you are also running a very veey involved OS that is constantly managing lots of services. Take for instance, more Ram is alloted jist to move the mouse cusor on a PC than all pf the OS memory reserve on the PS3/360. 

Another key factor is optimization. The word gets thrown around a lot but seldom ever truly understood. A dev working on a console will know that the console build he has for it will work on eveey single console of that platform. With a PC, devs will habe to code for things that they wouldnt even ever need to consider on a console (its kimda why throwing in a SSD in a PS4 wouldnt yeild the gains it would yeild on a PC. 

AMD and John Carmack said it himself that you can get out almost twice as much with a console than you can with a PC counterpart due to optimization. A 6tf console is equivalent to a 9tf = 10tf console easily.

No. A 6TF console doesnt just get 50-60% more power simply because its a console. You are taking what was said waaaay outta context. 

 No single  factor is responsible for the overall performance of any hardware. What is responsible for the performance of hardware is its system, how all its little factors work together. 

Teraflops is just a theoretical measure of a chips single and double precision performance. Key word being theoretical. But that performance is also toed to a lot of other factors. eg, you can habe the fastest GPU on tbe planet, but that GPU will only crunch data or map assets that it has recieved. This concenpt is how the term bottlenecks come about. 

The key advantage of PCs, is that they can brute force their way through anything. And this si achieved by putting higher end components in each area. Like throwing in an NVme SSD for instance, or even a better MB, or more powerful GPU.... etc.

A chip being on console will give you a 10-15% gain at most, and thats assumimg the chip doesmt have any kther bottlenecks to contend wtih. Take the XB1 for instance, the reason most games are sub 1080p isnt actually the fault of the GPU. Its the fault of using DDR3 ram and having on 32MB of "fast enough" ram. The reason most games on the PS4 don't hit 60fps isnt the fault of the GPU, its the fault of the jaguar CPU. It basically would hsbe to work twice as fast to hit 60fps.

I would disagree of the 10-15% gain only, OS footprint and overhead of DirectX or the likes would make that gain a little bigger.