They all have one thing in common. They are separate ISA's.
x86 does have an advantage where it is a CISC design, but RISC internally, so schemes such as this are actually one of x86's strong points.

The games aren't being recompiled, in-fact for some games that is an impossibility anyway as the source code is out of Microsoft's hands.
Rather... They are being repackaged, hence the required download.

What Microsoft has done is created a virtualized space, with emulated hardware pieces and using something very similar to binary translation to make ends meet.

And that is in conjunction with the fact that the Xbox One has native backwards compatible support for various Xbox 360 and Original Xbox technologies in hardware which makes backwards compatibility that little bit easier on hardware requirements.

Well that's just it.
If you were to throw double precision floating point, the Cell would not be breaking any speed records, unless you are comparing it against a Pentium 4. Jaguar can do 3flops per cycle for FP64 or 3 x 1,600 x 8 = 38.4 Gflop of FP64.
Cell is 20.8 Gflop. - 1.8 per SPE, 6.4 per PPE.

Which is why outside of a couple edge-cases... Jaguar beats Cell hands down, no contest. And to think it's a low-end, netbook CPU that is almost 5 years and was the worst x86 processor in AMD entire lineup of worst processors, but it still beats Cell.

The bulk of a modern x86 processor's transistor budget is actually not spent on processing capabilities itself, but rather on ways to keep the processor busy and hiding latency deficits.

We can throw all the cores we want at the problem... And those Cores can be as wide as you want... If they spend most of their time doing nothing, then it doesn't matter how fast the processor is, it's still doing nothing... And that is Cell in a bag.

GPU's are parallel processing monsters, but tend to lack on serial processing capability that a CPU is more suited for, which is why the CPU continues to exist today.
The CPU and GPU are both good at different things, but can both technically do the job of the other, they just happen to be bad at it.

This is just a bad idea in every sense of the word. The cell was powerful no doubt, but it was extremely difficult to use. And thats putting it mildly....

Then secondly, its like you have absolutely zero understanding of all the things that has made the PS4 successful. The PS4s success is clear and apparent that going with an "off the shelf" (even though thats a gross understatement of what these platform holders really do) solution was the better choice. Yet here you are suggesting that sony makes yet another over developed over complicated piece of hardware that will only be taken advantage of by their first party studios.

And this group think nonsense has to stop. Yes, the jaguar CPU used in the PS4/XB1 is weak by the general standards of everything else that is out there. But I can't stress it enough when I say that like most things that tech evangelists would like you to think is important; its really not important. Just look at the games that we have got this generation so far. Look at the games running on PCs with CPUs that are literally 10 times more powerful than what we have in these consoles..... are those games 10 times better?

Its flat out ignorant when people ask or want things simply because they can and not take into consideration all that really goes into these decisions.

And do you realize that if they change the core hardware architecture for the PS5 (meaning they heavily deviate from the current setup) then it means PS4 games won't work on that platform? Make no mistake, if sony or anyone releases a console in the future that isn't fully BC, that console will be DOA. We have the rise in digital distribution to thank for that.

Right now, its just better for sony to stick with AMD, make modifications to the chips AMD gives them like they did with the PS4pro and make the most they can of that hardware. Anything else at this point is just redundant and stupid.

Keep an eye out on checkerboard rendering..... its going to be a very big component in the next PS console hardware. Long and short is this, if you are looking back or suggesting taking a step back to make future improvements, then you already are doing it all wrong.

PS3 architecture was the biggest mistake of Sony. they learned nothing from the already complex PS2.

$50 is a massive cost to add to a product. It's not cost-efficient at all.

holy shit, it's virtually 2018 and someone is still wanting a PS3 cpu on an unreleased console?

this is either a very committed troll or just pure insanity :D

The only valid reason that I see would be for backwards compatibility, and even then I don't think it would be worth it. The PPE portion of the Cell can in no way compete with the Jaguar cores, and the SPE portion has been succeeded by general purpose programable GPUs (GPGPU)

1. Integer performance and IPC on Jaguar far surpasses the Cell. Jaguar also has much better power consumption as well.

2. The SPEs were designed for highly paralyzed operations, that is why certain tests might have cell coming up on top by a very marginal amount (and I would even question that marginal amount, because the test might have been better optimized for Cell than Jaguar). The problem is that modern programmable GPUs are far better optimized for those tasks than something like the SPEs ever where. Cell and its SPEs were designed in an era where GPUs were only starting to become more general purpose but were still very limited in how they could be programmed, as such highly parallelized operations would still run on CPUs (which were not designed for these types of operations and hence would run slowly) much of the time; Cell provided a stop-gap solution through the SPEs, but that solution is no longer necessary since such operations run way faster on the GPU.

3. Because the SPEs have been superseded by general purpose GPUs (GPGPUs) that portion (and according to you the key portion of the Cell) are no longer needed, which just leaves the woefully out of date PPE that has been surpassed by the Jaguar cores in every single way (look at point 1). The Cell provides absolutely no benefits in this day and age. Also, no programmer is going to take their parallelized code out of the GPU (where the code has most likely been optimized to run on multiple platforms and runs better than it ever could on a CPU or even something like the Cell's SPEs) and rewrite it so that it would run slower on the SPEs and no longer be applicable to any other platform.

Cant do it, because devs dont want to develope for the Cell.  It was the early downfall of the PS3, and sony wont risk going back to that well, nor should they.

PS5 has to be dev friendly, consumer friendly, and affordable, jut like the PS4.

They can but it's going to make NGPS more expensive which may slow down its adoption rate.they rather use that resource to actually make NGPS more powerful and use that power to soft emulate ps3 games.placing two CPUs inside a gaming console is not good for anything.imo they should make NGPS more affordable as they can.if it launches at 350$ ww then it's guaranteed that it's going to break some record.

Even then, the whole BC argument is stupid... A PS5 will easily be powerful enough to emulate the PS3 if Sony wants it (the XBO can, after all, emulate the 360. Now, while the PS3 is afaik a lot harder to emulate, the PS5 should also be a lot more powerful than the XBO...). Not to mention PS2 and PS1. That only leaves the PS4 which is easily achieved by sticking to x86.

I'm pretty sure there will be some SSD cache in next gen consoles, if not full SSDs. And they'll sell it as if it is brand new never before used technology.