Vor said: I mean, using an architecture that is hard to develop for while the competitor use the more standard and easier to develop architecture is better right? Now the developer will have even more excuse if the game isn't running smoothly

I am not arguing for the use of Cell, but it drives me crazy when people comments are way out of context.

The O.P. clearly said that the PS5 should have a Ryzen Based X86 CPU at the heart. So this would have no effect on backwards compatability.

Cell in this case is basically being used as a sub processor to handle things like physics, sound, and other CPU heavy effects. 

The O.P. is saying games would be built around the X86 Ryzen CPU, while Cell would work as a specialty processor to handle addtional tasks that would burden the main CPU.

Personally, I don't think it makes a lot of sense, as there are plenty of better, and more supported sub processors available. PS4 already uses multiple sub processors, and I would expect PS5 to do the same. Freeing up CPU cycles with speciallty processors is a huge benefit for consoles, but Cell is just not the right chip to handle these tasks.

Had Sony used the Cell in PS4 for sub processing/backwards compatability, and continued to invest in refining its achitecture and shinking its fab process then it would be a different story. Sony already ditched Cell, and it has not been refined or shrunk in years. It is extreamly unefficient and outdated tech at this point.

I don't think PS3 failed simply for price. The problem was it was $499-$599 and had little to show for it compared to 360.

Granted it had the Bluray drive and on paper the Cell was amazing, but the games were generally the same and sometimes worse due to optimization early on. Also, PS3 was sold at a huge loss so price cutting to compete with 360 was a financial disaster. That's not really reflective of today.

I think people who buy consoles at launch are willing to pay $499. With $50 price cuts every year. Also, a lot of people will stick with PS4 as the PS5 price drops. PS4 will be supported for years into the life of PS5.

Sony could make a decent machine for $399, but I think they can still thrive at $499 and it would make a big difference in specs. I think $499 is MS's aim now, so if Sony sticks with $399 its gonna be a pretty big technical disparity. Also, I really believe MS plans to support X1X into the 9th gen as budget hardware.

Neither one of you seem to understand that the Cell processor is Power architecture and thus, is inherently incompatible with x86

So the whole premise is nonsense. 

Right, maybe in the holiday of 2020 they can make a product that is considerably more powerful than the X1X at $399.

But will feel like a generational leap from the X1X? I kinda doubt it.

Basically this, or PS should have stuck with Cell for good as the main processor and kept upgrading it for PS4, PS5, etc. They didn't and chose the x86 route and will now stick with x86 for good until something way more powerful yet efficient comes along that isn't a pain to program for, which is a long ways off if it happens, if at all.

Adding Cell for BC is something I could maybe see in a Premium model on top of the Pro, but it would be extremely costly and may not sell all that well, in which case it's not really worth designing in the first place. I think this happening is unlikely.

Why would you alienate yourself from the majority? That would mean no PS4 BC for PS5 unless they work some magic.

Sorry mate, what you're proposing is beyond hilarious...please, don't dabble with topics you don't understand.

Your agreement isn't necessary.
The Cell is "hard to program for". - And there are a multitude of reasons for that relating to it's architectural hindrances. (Such as it's cache setup.)

No.

This has no relevance to the statement in my post. Try again.

Go do some basic research, gain an understanding of what flops is and how it relates to gaming and how irrelevant it is to the total performance of a system.
Then come back and try and have an intellectual debate about this particular topic.

No.
Go look up integer performance comparisons.
Again, you lack a basic fundamental understanding of how microprocessors work and their performance.

No.
The Cell lacks a ton of the basic functional components to perform GPU-type duties efficiently.

The Cell was designed to be a CPU first and foremost, the fact it is using the PowerPC ISA is a testament to that very basic fundamental fact... It sounds like you have been drinking some fake news over the years. - Of course they did take a VLIW-like approach to it's core layout with Cell. But ultimately... Who cares?

Do you know how rapidly GPU's and CPU's double their performance? Cell is ancient, outdated and downright archaic today.

If you think cherry picked benchmarks designed to conform to your own confirmation bias is somehow an accurate representation of Cell's capabilities... Then you are highly mistaken and you should again... Go do some basic research.

The Cell is great when doing Iterative refinement floating point, but you throw a ton of integer calculations at the chip and the Cell will fold.

To dumb it down for you (as it seems you need it) the Jaguar core is a more well-rounded balanced architecture that is essentially great at all sorts of calculations, where the Cell is only great at one thing and terrible at everything else.
Which is fine if you only intend to do a single type of calculation... But guess what? That isn't how game engines work, they use all sorts of calculations.

Besides.... The games speak for themselves... And the games have told us that Jaguar beats Cell. It really is that simple.
Playstation 4 games not only have more advanced A.I characters, but have more of them on screen.
Allot of multiplayer games (battlefield for example) have more players in a multiplayer match.
Also tend to have more impressive CPU based Physics, Particles and Smoke effects.

Clearly your knowledge on Ram is lacking. That is bandwidth per pin.
DDR2 can technically be faster than XDR2 if you take it wide enough.

However... You have conveniently omitted HBM, HBM2, GDDR6, GDDR5X as well in your comparisons. - So try again and try using sources that don't conform to your own confirmation bias.

XDR2 has been beaten soundly by other technologies.

Impressive for a console maybe.
But it paled in comparison to what Intel was offering.

AMD's GPU hardware isn't great compared to nVidia because AMD's GPU architecture is shit. -  It's more compute focused whilst nVidia's is a little more well-rounded and gaming-orientated.

In short, AMD GPU's can have tons and tons of Gflops... But games tend to need more than that, so AMD GPU's perform terrible, that's got nothing to do with developers or whatever other nVidia-having-a-monopoly conspiracy theory you have conjured up, that is strictly AMD's own fault.
It does mean that AMD's GPU's are great in tasks that can make use of all that compute like Bitcoin Mining, but not much outside of that.

Ryzen is actually a great CPU core, but it is limited by it's manufacturing, being stuck at an inferior 14nm Finfet Samsung-20nm-based Global Foundries process didn't really do it in any favors, especially when it comes to pushing up clock rates. (There may also be a bottleneck in say... The pipeline, not sure.)

We are not in 2005 anymore. x86 is extremely efficient, Intel even managed to get it's x86 processors to be competitive with ARM.

Citation Needed.

$800 was for the entire machine.
Throw in some overpriced Rambus DRAM, expensive Blu-Ray player, heck include a free PS2 inside every console, Hard Drive, memory card reader, graphics card, power supply, controller. HMMMM. I wonder how much is left for that CPU?

It certainly ain't approaching a several thousand dollar x86 chip that's for sure, let alone a first-gen i7 hex.

The Cell was cheap, it had to be, it had to go into a console.

The Cell's transistor counts are actually fairly low by modern standards and thus wouldn't be expensive to manufacture.

The real issue is that by opting for Cell you would need to take a multi-chip approach, which means implementing additional buses, controllers, caches and so on, which drives up costs... And if you drive up costs, that means you need to make a sacrifice somewhere else. - Hows about a cutback to the GPU?

Sony chose the right hardware at the right time for the right price for the Playstation 4 and they were rewarded that with sales.

What would be the point? Just add a couple x86 or ARM cores onto the SoC.
Cell isn't that powerful.

Besides, Physics and Sound can be done on the GPU anyway... And more efficiently than a CPU to boot.

Binary Translation.

For reason already mentioned, it would be an inconsequnt thing to do so. From an economical and development point of view.

Sony abandoned the Cell for good - not because of "se gud competition" as some falsly believe - mainly for the fast paced times we live in - where tech life cycles decrease steadily, new technology reaching the consumer market fast and prices decline quick, the consumer replacing old with new in shorter intervals - relying more on off the shelf solutions has become the wise thing to do. It is no coincidence all three console manufacturers have/had AMD as their supplier. Manufacturing capabilties, power consumption, versatility, cost efficiency ...

Research & develop complex hardware just isnt worth it anymore. It would be too short lived to justify it economically.

... Thats why we got the pro and the something X.

I am not even making a case for this. I know that Cell is Built on IBM's PowerPC tech. That said, there is no reason you couldn't have a device that uses X86, ARM, and PowerPC if you wanted. PS4 has an X86 chip and an ARM chip. They take care of different tasks. Sony could have used the Cell for these task, and offered B/C, but they chose the ARM chip instead. The PS5 will very likely have X86 and ARM, and additional chips to handle other specialized tasks as well. It is highly, highly unlikely that they would use Cell for these task in PS5.

As I said, I am not campaigning for Cell in PS5. People where commenting with points that had nothing to do with what the O.P. had proposed, and I wanted to address that. There are many, more powerful and effcient options to take care of anything Cell can do nowdays.