While I can certainly see his strict observation of transistor count doubling every 18 months reaching its limit (as it has already begun), I don't think it spells the end for improvements. People are just starting to get the hang of multiple cores after the ten or so years they've been on the market, and the architectures are improving quite a bit. Nevertheless, we still have some breathing room in terms of transistor size to push a little bit futher.

I'll admit that Geekbench doesn't really tell quantitative relationships very well, but it does do a decent job ordinally (i.e it tells which is better than which when they aren't close.) All of the benchmarks I have run on my various devices (phones, netbooks, laptops, gaming desktops, htpcs) seem to fit quite well with other documentation on relative CPU performance. At the very least the current Iphone has matched the low-end laptop CPU's. Furthermore, since we are talking about jaguar-based CPU's (in consoles) I see no problem with comparing it to broadwell, both of which are low-end, but the latter with a much better IPC. 

I think it will start out as (relatively) stagnant, but there is more money and effort than ever to find alternatives to die shrinks for performance improvements. And since Intel has a much smaller market control than in the past, there are more companies working on these alternatives and competing.

We already have had plenty of breakthroughs in the last five years or so. The only thing preventing their application is cost, but considering how everything has become much more mobile over time, there is a greater demand to solve the cost problem. I'd bet we'll see things like micro-supercapacitors entering mobile devices (or at least starting to) in the next ten years. That solves a lot of the heat problem as these capacitors have much greater power-densities than standard batteries, and can deal with higher temperatures better. When you can charge your power source in a matter of seconds energy density becomes much less of an issue. Even still, they are also moderately closing the energy density gap (see: below.) Just because the technology has remain stagnant for 25 years does not mean it will continue to remain so. There is a lot of money to be made in replacing the quite piss-poor battery technology we have currently, and tons of people trying to solve this problem on both the scientific and economic ends. I think it will happen very much like the transition from mechanical HDD's to SSD's (the first being a standard for decades as well, and is briskly being replaced as SSD's become cheaper.)

http://phys.org/news/2015-09-micro-supercapacitor-unmatched-energy-storage.html

or just a few days ago

http://arstechnica.co.uk/science/2016/02/high-energy-tiny-little-micro-supercapacitors-built-directly-on-a-chip/

And as I pointed out those very same chipsets that go into pico-itx are going into smart phones. The point is with many smart phones having HDMI out, you can keep your phone plugged in..the point is when they are talking about the smartphone chips that ARM holdings make it is about the chips they design , which does go into smart phones & tablets.

 "That means we are about a year away from having smartphone and tablets that are capable of running the same games"

Which is "capable" see the point is if developers & publishers take advantage of said hardware. That means new engines, middleware, resources. That does not take away from the fact it very well could.

The allwinner A80 is a complete joke. It is used by no major smartphone/tablet manufacturers and is design to be used in multimedia/cable boxes with active coolong and a plugged in power supply not mobile at all. Futher the A80 is less than a third as powerful as the PS3 dont know where your knowledge comes from.

Allwinner A80

• 68 gigaflops of single precision floating point performance
• 14.9 GB/s memory bandwidth
• A15 & A7 glued together in attempt to increase performance

It doesn't but it's the beginning of a new stagnation ...

@Bold It really doesn't. In fact the A9X (which is the bigger version of the A9 used in the iphone 6s) often gets hammered by a Core M. Linus Torvalds derides Geekbench for a very good reason ...

As per the last line I meant that they were targeting lower clocked (1.1/1.2GHz) Broadwell parts instead of comparing it to every variant that it has to offer at that power range ...

There's certainly more effort but we're nowhere near close to reaping the benefits of that investment. Transistors will be here for the next decade at the LEAST ...

@Bold It's not because cost is an issue, it's that no researchers in the laboratories has made any viable prototypes meant for daily use! Cost is a small issue as tech giants in the industry like to pick on low hanging fruits ... 

There's tons of technology that have stagnated for a long while. The engine that you see in your car isn't all that different from 5 decades ago in terms of concept. Your kitchen appliances haven't exactly advanced all that much and neither has your central heating system ... 

The transition from mechanical storage to solid state storage has been painful to say the least. I think you might be overestimating just how powerful we are at innovating ...

I don't give a fuck how powerful smartphones are until they stop becoming slow and glitchy after a few months of regular use. It's such a scam, these things are not built to last long.

There is really 2 issues here. Firstly these chips are at least 2 years away. Yes it is possible to produce chips with more power than an Xb1 and PS4 and they will probably achieve that in the next 2-3 years. But NO it won't be possible to produce those chips at that power and put them into a smartphone form factor, at least not unless you want to walk around with a portable liquid cooling system and have a battery life of less than an hour.

1. Off that was peak performance for the PS3

And 2nd and the snide comment you pointed out is not only incorrect,  its it off the performance index is the A80 is true 8 cores they are not glued together for as an attempt to increase performance it in fact does increase performance, the CPU along with the GPU are all inside stacked in the Soc. No south bridge, no off chip north, no daughter board.

 For instance

Ascl example of A80 boards usage of their powerVR GPU's in the SOC.

*

PowerVR G6230:  two Series6 USCs – 64 FP32 ALU cores with up to 128 FLOPs per cycle – 64 FP16 ALU cores with up to 192 FLOPs per cycle. That means up to 115.2 FP16 GFLOPS and up to 76.8 FP32 GFLOPS at 600MHz

So no, the PS3 CPU alone does not outperform the entire A80 boards.

Yes the PS3 GPU added in , yes, but this is about performance in how it can operate a year from now on mobile chips and their advancement is increasing fast enough that yes it will in short order even surpass the PS4 and Xbox one on a watt/performance scale yes, no doubt it would. It's not about "raw" performance. In AMD its always been about watt/performance on lower power ,Its based on Watt/performance under certain wattage for mobile's. Which the consoles & PC cannot compete with. Try a console power under 15 Watt's , how much performance could or would you expect? On that much power.

And in that case ARM is indeed correct.

Goal shifting? Neither ARM or Venture Beat in the OT nor anyone else in this thread were talking about performance per watt. Better raw performance than home consoles was the claim.

Read what he stated..again he talked about show "visuals" = & surpass consoles & PC's

Now again how is it goal shifting. These Single board chips can do visuals @ 1080P at voltages & performance in resolution that again, console like the PS4 or xbox one just could not compete @ ...the point is lower power/watt is the strong point ARM Holdings always concentrated on.

And what they pointed out is true. Could at that wattage games be made for Arm chips at such voltage show visuals = to the PS4 & Xbox one in a year from now? 

No its not, its a lie.

No mobile device (something you hold in your hand), will in 2017 be more powerfull than a PS4.

This is obvious a lie, its not technically possible, with the laws of physics and the technology we have now (or a year from now)

"that wattage" ?

You mean at 3watts like a smart phone? or 20watts like the Nvidia Shield TV console (which isnt a mobile device)?

Even at 20watts, in 2017 you wont see a "mobile device" anywhere near the performance of a PS4.

also your again changeing the goal posts.

Now your saying "equal" visuals (at much lower resolutions?) is equal to the PS4?

The orginal claim by ARM wasnt about visuals, it was about pure performance.

And its a strait up lie by ARM, so they make big headlines.

For some reason alot of people dont even question them, when they make rediculus claims.