I'm going to go out on a limb here and say that the PS4 will only be slightly more powerful than it's predecessor. I think it will be an optimized Cell or maybe even a dual-Cell but only if needed. I think that Sony has learned that being on the bleeding edge of technology is expensive and they will release a more cost efficent console to re-coup what is lost during the PS3 lifespan. I think it also depends on when the next gen starts and what the competition is bringing to the table. If this gen follows traditional 5-6 year gen shifts we could see the first next gen system in late 2010. Going by that timetable Sony would want the PS4 to be released close to that time if not soon after.

By that time devs will still not have fully mastered the Cell so Sony will stick with it and give the devs more time to sqeeze the full potential out of it.

ok I thought the PS3 was suposed to last 10 years ??? in my book and all the technological theories on increase of power... the cell is going to be outdated way before 10 years....

I don't remember who said he doesn't think the cell will be outrun in the past 2 or 3 years.... they probably have already in R&D centers prototypes or schemes of something better.... the development of the cell started in 2001 and took 4 years so I'm pretty sure since 2005 they have been already working on something else and in 2 or 3 years we'll find better or more convenient for video game consoles..

now beside that.... with the financial mess Sony is right now for the PS3 even if they come back and finish second they have very little chance of pulling out substantial benefits out of the PS3 by the time other like nintendo and probably MS will release a new gen.....
the real question should have been will sony be able to do a PS4.... and don't tell me it's the same for MS.... Sony has nowhere close of the financial backup that MS disposes...
that said it's the entire company that is going down financially lately with the gaming division leading the boat.... MS might be sinking a tone of money but last time I checked their total results were still up....
I'm not saying they won't have a PS4, and even less that they shouldn't have one.... I strongly believe it wouldn't help us customer to have sony out of the race... another monopoly on high def / consoles for MS is not a good thing for us... (I don't consider the WII being on the same market, not that it is better or not, nintendo themself think that way).... but I think that question deserves to be asked.... will sony be able or should stay next gen????
I'm asking the question on purely financial point of view.... no tech, no taste, no better or worse there involved....
is there anyone out there who think the question is valid and has maybe some substantial financial and economical facts and knowledge to sustain or refute the theory ???

(yeah its for my group project in financial economics :P )

Your entire post is kind of the point of this thread. If the Cell has a long enough lifespan to power the next PS, it's going to cost Sony very little to produce it. Slap two Cells in it, give it a gig or two of ram, and something fast for a video card, and your 90% done. All your PS3 development tools (and the knowledge learned this generation by developers) carry over.

This would allow Sony to release a very powerful and reliable console for $300 or less. Will MS be able to do the same thing? I think not.

Oh, and while Sony is no MS when it comes to making money, they did well as a company last year, and there gaming department is making money again. They are not in trouble or anything, they just made less then they would have hoped. 

I think people are missing the point of where Sony and MS have been headed with their architectures when thinking and talking about the next gen.  

MS went with a standard processor for the Xbox because it was quick and easy and for the sake of development tools.  They then migrated to PPC because the PPC platform scales extremely well, is cost effective and is perfect for doing emulation (backwards compatibility).

Sony spent megabucks developing the Cell because again, they wanted a PPC derivative platform that would scale well, etc.  Same benefits, different derivatives of the PPC architecture with the Cell being more specialized.  Whether that is truly BETTER or not remains to be seen.

In any case, I believe MS will come out with a larger version of the 360 basically.  Improved graphics card, maybe a physics chip, and more cores available of basically the same architecture.  Development tools would largely remain the same, not a TON to learn for developers, and they could do it pretty cheaply.  Going back to a standard Intel multi-core architecture would likely shred any plans for backwards compatibility (unless it's cheap enough by then to include the 360 processor in the box as well!) because of the fundamental differences between the Intel architecture and PPC.  Endian difference, and other architectural differences are just too large.  Same problem Apple had in switching to Intel from the PowerPC chips.

Sony had the Cell developed not just for their gaming division (although that was the biggest reason) but also to include in their other products as future needs dictate.  So while they invested a TON, the payoff is potentially there for them.  The next Playstation will likely use multiple Cell BE's with a graphics card.  This will allow them to leverage their past investment, lower costs and have a much more powerful platform as well.  In the end, the differences between MS and Sony are less although Sony will have more cost benefit since they've already paid to have the Cell developed.  As far as the difference in power, while most people acknowledge that the PS3 is the most powerful console overall, it remains to be seen how that will translate to software.  So, multiple Cells may give them an edge and it may not.  Either way, MS has a way to easily produce a comparable console.  And so the war will continue. 

IBM already has monster chips that dominate Cell in servers. I forgot the name but they are freaking monsters. I believe they are over 4 GHZ as well.

You dont understand technology well, like most posters on this site you're probably just a little kid. Theoretical flops from Cell dont necessarily mean anything in the real world. Because what happens as soon as you throw a branch at it? A lot of gaming code involves branches where Cell kind of sucks. Hell, John Carmack thinks X360 CPU is better than Cell, and I promise he knows a lot more about programming than you. Cell is one of the main reasons PS3 is difficult to program for and why so many PS3 ports dont run as fast or look as good as their 360 counterparts.

I hope Cell(s) is in PS4...because I dont think it's a very good chip, quite frankly. I think that will give MS an advantage in the next generation. About the best that can be said for Cell is it is a good CPU at helping the GPU out. The problem is, you are better off spending money on simply making a better GPU in the first place, because anything a GPU is good at it will do 10X faster than Cell or any CPU.
If Sony had spent half the money they spent on Cell on a better GPU for PS3 instead, they would't be in a situation where 360 games look just as good.

Ask yourself which is better, Cell+RSX or, Core2Dou+8800GTX. The latter can run Crysis. Go download some Crysis vids or better yet find a way to watch it yourself on high settings. It simply destroys anything on consoles including Uncharted. Why? Because you are better off with a great GPU and a medium CPU than a great CPU and a medium GPU every time, at least when it comes to graphics.

BTW, John Carmack said PC CPU's will be twice as fast as the current console CPU's like two years ago. You aren't looking at things like a programmer. Cell in certain trivialized applications is really good at math, it struggles in a lot of other places.  A core2dou is three or four times as fast as Cell andXcpu in reality, from a programmer perspective.

@sharky,

You have no clue what you are talking about. Sorry, I don't usually insult people on here, but when it's due, it's due.

Do a Google search on the CBE, and look at what researchers are able to do with it. I have linked, more than once, to real world research, and what I get back from you is that “I must be a kid”, and second hand comments from a John Carmack interview.

You said “Ask yourself which is better, Cell+RSX or, Core2Dou+8800GTX.” Well, depending what you are doing the Cell can do it many times faster than the Core2Dou+8800GTX alone, forget the RSX. Can it do everything better? Of course not, it’s a specialized CPU, but what it specializes in is good for gaming.

Ask yourself this. Can you sell a Core2Dou+8800GTX for $399 and stuff it in a small box with little heat displacement? Also, if you use your system 3 hours a day, how long will it last? How much power will you use at the end of the year (and we are talking about $50-$100 a year more, it’s not trivial).

You can’t just look at numbers and assume it’s the next gaming system. The Cell is incredibly powerful for what it’s good at. As the poster above you stated, we will see how well it ends up working out for games.

Real world research on what? If you make a trivial aplication, and code it to suit Cell, make sure all your data fits in the tiny local store and just iterates, yeah it can look great. It can crunch a lot of flops. That's meaningless in the real world. Ask Carmack what he thinks of flops.

You dont even have the slightest idea of programming and dont know where the Cell is weak-branching code and things like that. You are the gullible person Sony marketing aims at, taking meaningless flops at face value.

Cell hasn't made any difference where a CPU is supposed to, animation, physics, AI..you cannot tell any difference between a 360 game and a PS3 game in those areas. The best physics I have seen are in Star Wars Force Unleashed or maybe Orange Box..they are both multi-platform so they're on 360 too.

Cell is a very big, hot chip comparatively by the way. It's die size (you probably dont even know what that is) is quite a bit larger than Xcpu which means it's more expensive to produce and it uses more power. Blu Ray is the biggest reason PS3 is so expensive, but Cell is a reason as well. The die size of Cell is 234 mm^2 and Xcpu is 160 mm^2. Chips cost according to die size so that means Sony is paying more for Cell than MS is for Xcpu, not too mention Sony spent hundreds of millions on Cell R&D. That's fine but the question is "is it worth it?". IMO the answer is clearly no. They would have been better off using a cheaper CPU (but just as effective, like Xcpu) and adding somewhere else with the money saved, such as a better GPU or more RAM in PS3.

It's exactly like I said already. Cell doesn't help with AI, physics, or animation like it is supposed to. What it is seeming to be actually being effective as in the real world is assisting the GPU. That's ok, but Sony would have been far more cost effective at spending more money on the GPU than asking the CPU to do stuff it isn't designed for and is slow at.

BTW, yeah, a 8800GTX would fit fine and be reliable in a console. Something far bigger and more powerful than a 8800GTX will be in Xbox 720/PS4 anyway, so how do you think 8800GTX wont work?? And yeah, there are 8800's in laptops..power, it probably draws less than RSX. RSX is just a 7800GTX. 

sharky,

Don't waste your time, TheRealMafoo doesn't understand what he is talking about; he doesn't know the difference between Ray-Tracing and Ray-Casting, and thinks that the cell processor is a super-computer because it is more powerful than a 5 year old processor. He will continue to post links to articles which compare the Cell to outdated processors, or praise its performance in algorithms (like the Fast Fourier Transform) which has absolutely nothing to do with gaming.

He probably doesn't understand the challenges associated with load balancing, concurrency, and the general inefficiency that comes from parallel processing. He can't tell you how implicit solutions to differential equations relates to a physics engine, how a complex number represents a rotation in 2 dimensions (which is the simpler explaination as to why a quaternion represents a rotation in 3 dimensions), or how you can represent a bidirectional reflection map (a 4 dimensional texture) as 2 2D textures to create impressive material effects on modern (and now outdated) GPUs ...

He is a poser ... 

Sharky just grinded that guy's face into the ground.

Well to be honest, I was reading something about somebody working on something very similar to cell except 4 actual cores with 25 "Cells" per core with a high clock.... last year. Of course something like this would probably be military classified only until like 2010+ lol I'll try to scrounge up the article.