Support NeoGAF

[cyberheater]

Not even close.

IBM and Sony are both still putting money/research into it. IBM has already hinted that the Wii U is using their latest CELL processor.

WiiU processor is based on POWER7 tech. That's not the same as Cell.

 

[DCKing]

Well the PPE in the cell would most likely be replaced with a POWER/PPC core or two. My argument mostly pertains to the fact that going for GPGPU cores would be counter-productive considering that spes can still do there job better and those transistors could go to way better uses. Basically you get the benefits of OOOE cores with proper branch prediction and the raw performance of the SPUs with a capable GPU to free up its resources. None of that would break sonys bank either.

How would those SPUs be put to use at all next to a modern GPU? A GPU which can do the effects the SPU is used for now much faster than the SPUs can? I don't see the point.

Also im pretty sure bulldozers disappointing performance has turned off any interest it had in this field. I don't see anybody ditching IBM's CPUs.

Both POWER7 and Bulldozer run circles around the PPE. Circles and circles.

Those are usually protected by concurrency control mechanisms like barriers and semaphores. They still can be translated statically.

No, this is wrong. Barriers and semaphores are dependent on runtime conditions that cannot be preprocessed. The preconditions for that code is wholly undeterministic.

 

[dragonelite]

I think IBM said development would continue in the context of hybrid designs and next-gen architectures that carry some cell flavourings.

Honestly don't see what would be so outlandish or undesirable about a 'beefed up' Cell design - or - for example, a hybrid design with a few larger, newer, cores and an array of SPU compliant processors. 6 SPUs, with some tweaks, would be a fantastic for BC (obviously) and could do their bit for native PS4 applications alongside other main cores - and I have to guess would take up relatively little die area in 2013 (?) They could probably clock them quite substantially higher too, for the benefit of PS4 code.

I have a feeling, though, that the new systems will have a higher focus on GPU processing power, a higher ratio of gpu:cpu than the current systems did. There may be relatively 'small' CPUs in the next systems.

Kinda like larrabee was considered dead but the tech will life on in knight corner(??).

 

[KiNeSiS]

Gen almost over....

= (

 

[weekend_warrior]

If you have to spend a shed load of cash designing a new cell then it takes a long time to claw the money back. Which means a higher cost console.
If you use off the shelf parts then upfront dev cost are considerably cheaper.

They've already spent the money on CELL though, so unless they've have to dump a ton more money into it for PS4, then it shouldn't affect the price like it did for PS3.

They could double the amount of SPEs, clock at 4.8GHz and get great CPU with virtually no R&D.

Isn't that one of the big selling points of CELL, how easy it is to chain them together for significantly higher performance?

 

[gofreak]

How would those SPUs be put to use at all next to a modern GPU? A GPU which can do the effects the SPU is used for now much faster than the SPUs can? I don't see the point.

Spare the GPU to let it do other things, perhaps?

It doesn't particularly matter always if the GPU could do the same work with greater efficiency or power efficiency - it can still be a win to do processing somewhere else if it let's you focus more GPU power on a bigger bottleneck elsewhere. I've seen it talked about more than once where devs chose to put certain algorithms on the SPUs even though they were slower than doing the same on a PPE or the GPU - just because it freed up those other resources to fry bigger fish that was bounding the overall performance, and so lowered the frame latency overall.

A lot of devs might not bother just because there may be so much power elsewhere that it's not necessary, but among devs who wanted to squeeze every last drop out, they'd find a use for them, I'm sure.

Not to mention those resources could also be applied in other ways - for example for OS tasks. Maybe the small SPU array could be used just by the in-game OS.

It would cost Sony some die area to include them, but they wouldn't go to waste at all I don't think, and if Sony was happy to go with a bigger die than their competitors, it wouldn't be the first time. It could be worth it...they'd be an awful lot more useful than the chips that have been needed for BC in past systems

(This is all assuming they wanted to go with a completely different/new architecture for the rest of the CPU, and SPU or SPU compliant design wasn't on the cards for it...)

 

[iamshadowlark]

How would those SPUs be put to use at all next to a modern GPU? A GPU which can do the effects the SPU is used for now much faster than the SPUs can? I don't see the point.

You seem to only be thinking of the SPU as a mechanism to augment vertex processing and such like it is used today. The original SPU's are still to this day top of the line at physics and simulations. You can also use them for more post-processing power and additional vertex work as you can never have too many processing units. As I said there are still a wealth of things that you would rather have running on SPUs than on a GPGPU.

Both POWER7 and Bulldozer run circles around the PPE. Circles and circles.

Its odd that you would compare 2011/2012 tech to 2005/2006 tech. Especially considering I said the PPE would be nixed anyway.

 

[patsu]

No, this is wrong. Barriers and semaphores are dependent on runtime conditions that cannot be preprocessed. The preconditions for that code is wholly undeterministic.

Those are only a system call or CPU instructions. You replace the old call with a new suitable one. They won't make static compilation impossible. They will run as long as the original code is safe. However, timing may be more tricky to guarantee.

 

[REMEMBER CITADEL]

MS seems to want the new Xbox being a machine that can do everything but nothing great, and on top force feeding their Kinect technology to the customer. That would explain the rumored not so impressiv xbox 720 specs.

You, sir, win the thread, and in one sentence no less.

 

[DCKing]

It doesn't particularly matter always if the GPU could do the same work with greater efficiency or power efficiency - it can still be a win to do processing somewhere else if it let's you focus more GPU power on a bigger bottleneck elsewhere. I've seen it talked about more than once where devs chose to put certain algorithms on the SPUs even though they were slower than doing the same on a PPE or the GPU - just because it freed up those other resources to fry bigger fish that was bounding the overall performance, and so lowered the frame latency overall.

This reiterates what I've read and illustrates what's wrong with Cell. There's much power in the SPEs but no real use for them, meaning they are free to do the jobs that is meant for other hardware, even if it less effective at doing that job. If you have to write your software like that (and, importantly don't get much better results than the competition) you have badly designed hardware.

The graphics tasks Cell has been doing now are much better handled by a GPU, especially a modern GPU unlike what's in the PS3. That's why I don't see the point of Sony going with Cell as their main CPU again. That silicon is much better spent at the GPU and CPU cores.

Those are only a system call or CPU instructions. You replace the old call with a new suitable one. They won't make static compilation impossible. They will run as long as the original code is safe. However, timing may be more tricky to guarantee.

You don't understand what I'm saying. The specific instructions don't matter. It matters when they are called for correctness of execution. When they are called is dependent on runtime conditions which are unpredictable. This means that you can't preprocess code to make it emulatable on new systems. I'm afraid I can't explain this well enough over an internet forum, however.

 

[gofreak]

This reiterates what I've read and illustrates what's wrong with Cell. There's much power in the SPEs but no real use for them, meaning they are free to do the jobs that is meant for other hardware, even if it less effective at doing that job. If you have to write your software like that (and, importantly don't get much better results than the competition) you have ineffectively designed hardware.

That wasn't true in a PS3 context. I've seen a tonne of algorithms and comparative benchmarks presented at conferences where the SPUs were doing all kinds of things faster than RSX.

My point is just that even if in a next-gen context that were less often, or never the case, it wouldn't necessarily mean they would be 'wasted'. They'd still be put to use, I think, even if they couldn't be amazingly gracefully integrated with the rest of the design. Either by native code, or the OS. The (small?) die cost could be well worth the utility of BC, and it would be less wasteful than some of the silicon included in past systems for BC.

The suggestion, by the way, I don't think it was that they'd be the dominant feature, but added 'on the side' of a different CPU setup. A hybrid. Even if the design was closer to the original Cell design, though, I don't know if you have to worry about Sony spending too much on the CPU at the expense of the GPU - I've a feeling whatever the CPU looks like, it'll probably be dwarfed by the GPUs in the next systems, compared to the current gen. I think we'll see a much heavier lean on GPU processing by everyone, relatively. But that is a pure guess

 

[Sushen]

Seriously? Are we going through this again? With two vaporwares? OMG.

 

[patsu]

You don't understand what I'm saying. The specific instructions don't matter. It matters when they are called for correctness of execution. When they are called is dependent on runtime conditions which are unpredictable. This means that you can't preprocess code to make it emulatable on new systems. I'm afraid I can't explain this well enough over an internet forum, however.

Those concurrency control mechanisms just make unpredictable runtime conditions predictable. You can translate them from one form to another statically or even by hand to see that the new form can work as long as the old one is safe.

Data dependencies are harder to solve because of performance characteristic differences. They may need to unroll loops if they are too different.

 

[DCKing]

You seem to only be thinking of the SPU as a mechanism to augment vertex processing and such like it is used today. The original SPU's are still to this day top of the line at physics and simulations. You can also use them for more post-processing power and additional vertex work as you can never have too many processing units. As I said there are still a wealth of things that you would rather have running on SPUs than on a GPGPU.

So although it is quite likely the SPUs will be obsoleted by a modern GPU in the vertex processing area, you think Sony should keep them for their physics simulations? Keeping 7 SPUs for mostly physics, which is processing that gets by just fine on non-Cell platforms, doesn't sound like a very compelling design argument for a new console to me.

Also, I recognize the Cell's potential for post processing, read my earlier posts. It doesn't have to be the main CPU for that. Nor would it need the expensive modernization operation.

Those concurrency control mechanisms just make unpredictable runtime conditions predictable. You can translate them from one form to another statically or even by hand to see that the new form can work as long as the old one is safe.

Data dependencies are harder to solve because of performance characteristic differences. They may need to unroll loops if they are too different.

I'm not familiar with the restraints that exist for SPU code. For general purpose code however, it is in fact an impossible task to do what you say.

 

[gofreak]

Layman speculation time:

"I think you'll see [Cell] integrated into our future Power road map. That's the way to think about it as opposed to a separate line -- it'll just get integrated into the next line of things that we do," Menon said.

http://www.networkworld.com/news/2010/100810-cell-processor-development-hasnt-stalled.html

What they're doing 'next' is Power8.

Power8 is rumoured to be scheduled for 2013, and one bulletpoint feature is 'accelerators'.

In hybrid supercomputer designs, IBM has paired Opteron processors with Cell processors, the latter acting as...well...accelerators.

Maybe a Power8 - or perhaps even a Power7 - variant could use accelerators that are instruction and performance compatible with SPUs? It could be a neat solution for Sony if it was possible.

 

[bigtroyjon]

So I've been going back and looking at old info from before the 360/PS3 came out and man, E32005 where Sony was going on about the GFLOPS performance advantage was hilarious and this Gamespot article was full of all kinds of PR bullshit from that E3. This is my favorite in retrospect

In what must come as a relief to developers, Epic Games' Tim Sweeney was on hand to vouch for the PS3, saying it was "easy to program for" and that Epic had received its first PS3 hardware two months ago.

Looking forward to getting the same kind of laughs from MS and Sony again.

 

[patsu]

I'm not familiar with the restraints that exist for SPU code. For general purpose code however, it is in fact an impossible task to do what you say.

Statistically translating a thread safe code from one CPU to another is doable. Otherwise compilers won't be able to work correctly when it translate human readable code to machine code.

 

[cyberheater]

That wasn't true in a PS3 context. I've seen a tonne of algorithms and comparative benchmarks presented at conferences where the SPUs were doing all kinds of things faster than RSX.

My point is just that even if in a next-gen context that were less often, or never the case, it wouldn't necessarily mean they would be 'wasted'. They'd still be put to use, I think, even if they couldn't be amazingly gracefully integrated with the rest of the design. Either by native code, or the OS. The (small?) die cost could be well worth the utility of BC, and it would be less wasteful than some of the silicon included in past systems for BC.

That's quite a stretch. I'm sure technically it might be possible but it would add a relatively large and unnecessary cost in terms of hardware, software and crucially time. Sony do not want to be late next gen. Cutting the requirement for BC would be a easy low hanging fruit.

 

[rdrr gnr]

Looking forward to getting the same kind of laughs from MS and Sony again.

Aside from the unyielding fanboys and the mentally inept, I'd say everyone's bullshit detector is finely tuned. Once actual 'specs' come rolling in, I don't except it to be as insane as the current generation.

 

[H_Prestige]

That's quite a stretch. I'm sure technically it might be possible but it would add a relatively large and unnecessary cost in terms of hardware, software and crucially time. Sony do not want to be late next gen. Cutting the requirement for BC would be a easy low hanging fruit.

Cutting BC is the equivalent of pressing the delete button for the entire PSN store. Not going to happen.

 

[DCKing]

Statistically translating a thread safe code from one CPU to another is doable. Otherwise compilers won't be able to work correctly when it translate human readable code to machine code.

Again, I'm not talking about translation. And I'm not talking about thread safety. You will get different runtime behaviour (behaviour being different results, without violating thread safety), which can lead to nonsensical output and crashes. You cannot execute Cell code without emulating Cell's runtime behaviour accurately, which is impossible currently on non-Cell hardware.

 

[gofreak]

That's quite a stretch. I'm sure technically it might be possible but it would add a relatively large and unnecessary cost in terms of hardware, software and crucially time. Sony do not want to be late next gen. Current the requirement for BC would be a easy low hanging fruit.

It depends what they had planned. If it was a late requirement, yes, if not, I'm not sure why it would hold anything up.

In terms of software cost, I don't see why. Hardware cost? Yes, if it was an 'extra' on the side, but I think it could be a pretty small extra to have 6 on 28 or 22nm - and may be worth it for the utility of BC.

Anyway, there may also be more graceful ways to integrate SPU compatibility more closely into a new design. There have been hints that Cell work or aspects of the architecture may merge into the next-gen Power line...and maybe that'll include the bits that would be hard BC-wise, or bits that would more easily accommodate the customisations of a client...

 

[cyberheater]

Cutting BC is the equivalent of pressing the delete button for the entire PSN store. Not going to happen.

That doesn't make sense. It's not like PS3 is going away anytime soon. By the time PS4 launches, they'll be upwards of 60 million PS3 out there. Stuff will still sell in the store.

 

[StuBurns]

If Sony makes a clear display that BC is not something they care about for users going forward, and MS shows they do, it will be a significant issue for Sony. The further we go down the DD thing, the more important BC is going to be.

 

[H_Prestige]

That doesn't make sense. It's not like PS3 is going away anytime soon. By the time PS4 launches, they'll be upwards of 60 million PS3 out there. Stuff will still sell in the store.

All that content can only be sold to legacy users in that case. With BC they can sell to everyone, and I'd imagine the potential revenue there is a lot more than whatever money they "lose" from PS4 owners buying PS3 retail game discs instead of PS4 games.

Sony knows this. Why else is the Vita BC with PSP?

 

[cyberheater]

It depends what they had planned. If it was a late requirement, yes, if not, I'm not sure why it would hold anything up.

In terms of software cost, I don't see why. Hardware cost? Yes, if it was an 'extra' on the side, but I think it could be a pretty small extra to have 6 on 28 or 22nm - and may be worth it for the utility of BC.

The software cost would be incurred by developing a PS3 os emulation layer that sits on top of a PS4 layer that worked for a majority of games and peripherals.

Anyway, there may also be more graceful ways to integrate SPU compatibility more closely into a new design. There have been hints that Cell work or aspects of the architecture may merge into the next-gen Power line...and maybe that'll include the bits that would be hard BC-wise, or bits that would more easily accommodate the customisations of a client...

Given how hard devs are pushing the whole PS3 architecture. I can't see how you could emulate it accurately enough at the speed required even if you had some Cell in there.

 

[bigtroyjon]

Aside from the unyielding fanboys and the mentally inept, I'd say everyone's bullshit detector is finely tuned. Once actual 'specs' come rolling in, I don't except it to be as insane as the current generation.

Product launches will always be filled with PR bullshit and outside of a few people every thread on GAF suggests that most of us don't have a clue as to what the actual "specs" mean when it comes to gaming performance, especially for a specialized product like gaming consoles where pure specs don't really mean much with the specially designed components.

 

[cyberheater]

If Sony makes a clear display that BC is not something they care about for users going forward, and MS shows they do, it will be a significant issue for Sony. The further we go down the DD thing, the more important BC is going to be.

I just don't see why anyone would think that Sony has any interest in BC what so ever given that they ditched it as fast as possible on PS3.

 

[Aizu_Itsuko]

Let's not forget there were rumors of them using multiple cells instead of a GPU.

It's been a while but I think that was never the case, at least not just two identical Cells. IIRC Toshiba had some GPU part but in the end they went with Nvidia for some reason (performance, ease of use, cost, whatever).

Anyway, in the end the PS3 seems like a powerful but unbalanced console. As far as overall performance goes it seems to have an edge over the 360 but only if you specifically take advantage of the system, Cell included. When it's getting ports from other systems it ends up just being about the same as a 360 and depending on the port/game it can even lag a little behind. Things seem to change when PS3 is the lead platform but most of the time there's just parity.

Thankfully the R&D and manufacturing costs of Blu-Ray and Cell are a thing of the past and I'm sure the PS4 will be priced more competitively (and not be a money pit for Sony). MS won't go crazy with the price either. Same for Nintendo (even though Wii U could be more pricey than previous Nintendo systems).

As for this rumor, it's just that. It's not that I find it unbelievable but there's no point in discussing it until there's more solid information available.

 

[StuBurns]

I just don't see why anyone would think that Sony has any interest in BC what so ever given that they ditched it as fast as possible on PS3.

They ditched it as a last resort, they already removed half the chipset, going semi-software BC before taking the other half. They didn't want to do it. And they're doing it with PSV, a platform I don't think people really even expected to have BC. They cared about it so much with PSV they picked the resolution of the screen based on providing per pixel mapping for BC. Something way beyond the call of the duty really.

I think if there is any possible way they can include PS3 BC, it'll be in there.

 

[AngelSoldier]

I just don't see why anyone would think that Sony has any interest in BC what so ever given that they ditched it as fast as possible on PS3.

Yeah, but now it's coming back with dem' dare' PS2 games on the PSN Store.

 

[Aizu_Itsuko]

I just don't see why anyone would think that Sony has any interest in BC what so ever given that they ditched it as fast as possible on PS3.

Because they have always had an interest in BC.

They ditched it in the case of the PS3 because they had to cut costs fast as they were bleeding money and losing sales due to the high price.

Not saying PS4 will be BC with PS3 (that will depend on the costs) but they are interested in BC and it would be a nice bonus that would help sales.

 

[gofreak]

The software cost would be incurred by developing a PS3 os emulation layer that sits on top of a PS4 layer that worked for a majority of games and peripherals.



Given how hard devs are pushing the whole PS3 architecture. I can't see how you could emulate it accurately enough at the speed required even if you had some Cell in there.

I'm not envisaging emulation. I think the suggestion was literally putting some SPUs in the CPU (on the premise that 6 of them in 2013 would be relatively cheap wrt die area), or something that was SPU compliant. If you had that you wouldn't have to emulate. PPE code would probably run pretty OK on a regular Power too, but if emulation was necessary there it would be more doable in that case.

That's just a suggestion though. But I think if Sony went to IBM and said 'we want a next-gen Power, and we want to be able to run SPU code somewhere within that', IBM could make it happen. Who knows - maybe that's what Power8 is, next-gen Power + accelerators that either are, or can be made on a client's request to be compatible with the SPU instruction set. IBM has hinted at cell being folded into their next gen Power - whatever that ends up meaning. And they have got a thing for heterogeneous acceleration, at one or other level, and have mentioned accelerators as a new feature point for Power8.

This is all assuming BC is a hard requirement for Sony. It may not be at all. They were rumoured to be researching Larrabee early on, for example, which would suggest an open-mindedness on that front. But I don't see why 'Cell' has to be some dirty word if compliance becomes a requirement. I don't know how close the relationship is between IBM's next main line and cell is, but what relationship is there may make it more feasible or attractive to come up with a new solution that is also BC (or can be with some customisation) than people think. Or not We'll see.

 

[patsu]

Again, I'm not talking about translation. And I'm not talking about thread safety. You will get different runtime behaviour (behaviour being different results, without violating thread safety), which can lead to nonsensical output and crashes. You cannot execute Cell code without emulating Cell's runtime behaviour accurately, which is impossible currently on non-Cell hardware.

Naturally, an emulator needs to emulate Cell's runtime behavior accurately but using different CPU. Assuming it translates correctly but crashes, it would usually mean that the original program is not thread safe or synchronzised properly or has a bug, perhaps crashing occasionally even on PS3. They will have to tweak the timing in a case by case manner.

On PS3, we can replace the HDD with different specs, it should not crash because of load speed differences too. Synchronization between Cell input/output with RSX is usually controlled by a concurrency control mechansim (to indicate that buffer is ready). Those may require tweaking after automated translation if the original code makes too much assumptions.

 

[Marco1]

If Sony makes a clear display that BC is not something they care about for users going forward, and MS shows they do, it will be a significant issue for Sony. The further we go down the DD thing, the more important BC is going to be.

I agree and I think this is were MS will have sony and why I think halo4 will be sold on 360 and nextbox and also play multi-player with cross compatibility.
To prove that everything can transfer over easily.
I can see the top 10 live activity for the 360 being the same on nextbox for at least 1 year.
Edit: I read above that vita is BC for the same reason, I think Vita should have come with a voucher that offered 3-4 free PSP games on the PSN store.

 

[Metalmurphy]

I just don't see why anyone would think that Sony has any interest in BC what so ever given that they ditched it as fast as possible on PS3.

They ditched it because of all the money and sales they were losing. The PS2 hardware on the PS3 was never or very slowly going down in price and they needed to cut the production costs asap. Also the VITA having PSP BC means they do still have an interest.


But yeah, the only way BC is happening is if they use the Cell again so they wouldn't need two different types of hardware in the same machine.

 

[cyberheater]

They ditched it as a last resort, they already removed half the chipset, going semi-software BC before taking the other half. They didn't want to do it. And they're doing it with PSV, a platform I don't think people really even expected to have BC. They cared about it so much with PSV they picked the resolution of the screen based on providing per pixel mapping for BC. Something way beyond the call of the duty really.

I think if there is any possible way they can include PS3 BC, it'll be in there.

I know what you're saying but given how cash strapped Sony are right now and time to market for PS4. Add in the fact that when PS3 launched, there was over 100 million PS2 out there so it made some sense back then to add in BC but when PS4 launches, there won't be anywhere nearly so many PS3's. I just don't see it happening

 

[Metalmurphy]

I know what you're saying but given how cash strapped Sony are right now and time to market for PS4. Add in the fact that when PS3 launched, there was over 100 million PS2 out there so it made some sense back then to add in BC but when PS4 launches, there won't be anywhere nearly so many PS3's. I just don't see it happening

There will be about the same PS3s as PSPs, if not more. And there will certainly be alot more PS3 games then PSP ones.

 

[gofreak]

Some speculation by others re. Power8 and how Cell might be folded in.

http://www.h-online.com/newsticker/...s-About-Chancellors-and-3D-Chips-1206912.html

While these are only test chips, IBM intends to produce the Power8 processor, planned for 2013 (in 28 or 22nm), using 3D stack technology, with a directly linked memory and also, probably, a layer of small specialised computing cores adapted for specific intended uses. It is here that the successors of the Synergistic Processing Units of the Cell processors might turn up; the 3D stacks offer lots of room for modularity.

So maybe 'accelerators' in Power8 can perhaps be whatever a client wants, within reason - can be dropped in, dropped out relatively easily. Maybe the default configuration will come with something that's SPU compliant (e.g. a next gen SPU per the above speculation), or if not, a good old fashioned SPU could be put in there.

Dunno though, maybe Power8 will be a bit too leading edge for PS4. Kind of a shame if it was able to offer a neat solution but was too late.

 

[cyberheater]

I'm not envisaging emulation. I think the suggestion was literally putting some SPUs in the CPU (on the premise that 6 of them in 2013 would be relatively cheap wrt die area), or something that was SPU compliant. If you had that you wouldn't have to emulate. PPE code would probably run pretty OK on a regular Power too, but if emulation was necessary there it would be more doable in that case. etc...

But the next gen cpu would be running at a different clock rate, the memory speed would be different and so would the architecture etc... of the GPU. For it to run PS3 games, there would still be a high level of emulation or on the fly instruction refactoring and it would have to be near perfect. I would be seriously impressed if they managed to do it.

It might be possible to recompile the game code but still use all the existing data but that would have to handled on a game by game bases and you'd have to download the main code to your machine.

 

[NemesisPrime]

How much will Cell's cost in two years from now? Sony don't have to reuse it for PS4 itself just to include it if it's cheap.

True... that would be their best choice. Add an old Cell for BC and also let it do some work in PS4 games. We will see... I doubt they will do this though.

 

[H_Prestige]

I know what you're saying but given how cash strapped Sony are right now and time to market for PS4. Add in the fact that when PS3 launched, there was over 100 million PS2 out there so it made some sense back then to add in BC but when PS4 launches, there won't be anywhere nearly so many PS3's. I just don't see it happening

How much less does a non BC ps4 cost than a BC ps4? Why does it have to be handled by putting the ps3 cpu/gpu combo on board? MS sure as hell isn't going to do BC that way so I don't see why Sony would either.

 

[Marco1]

Will the RSX not also bring its own problems regarding BC if they don't use Nvidea again?
Are Nvidea GPU's not power hungry heat monsters that they will be hoping to stay away from?

 

[gofreak]

But the next gen cpu would be running at a different clock rate, the memory speed would be different and so would the architecture etc... of the GPU.

Heh, well, one thing at a time re. the GPU

Re. the clock rate, components can be clocked differently under different usage. Re. memory speed, I HAVE been conveniently ignoring I/O, but assuming a new chip was a superset of the old here in all respects wrt memory, a vm could be built to match its characteristics I think. I know this is easier said than done and all that, but if it was a requirement upfront, I think it could be done.

 

[StuBurns]

Will the RSX not also bring its own problems regarding BC if they don't use Nvidea again?
Are Nvidea GPU's not power hungry heat monsters that they will be hoping to stay away from?

Maybe if Nvidia really are providing something for a console, it's a variation on the RSX dedicated for physics simulation for PS4 and PS3 BC, with AMD providing the main GPU.

 

[Alebrije]

Sony needed to BC the PS3 because the huge amouth of great exclusive games that PS2 had , just the same with the PS1 to PS2 era. But now there are just few great PS3 games that someone could want to play on a "PS4".

So maybe it could be better to translate those few games to the PS4 just as the HD series we got on the PS3 from PS2/PSP games ( Sly, Ico, MGS, Tomb Rider, etc) and give the PS4 a next gen CPU and architecture decicated for its games.

Obviosly they need to give a 20 or more titles library at launch with 3-4 AAA games ...

 

[i-Lo]

If Sony had ditched the bluray, backward compatibility hardware and had a cheaper, smaller HDD and have launched at $399 then I think they would have got a lot more sales then including an ultimately doomed media format.

Don't know what you're smoking, but I'd like some of that.

As much it hampered the launch window for Sony, BD is one of the most distinguishing features of the PS3 that separated it from the competition and given that it was the cheapest BD player in the market helped it with the movie watching crowd. Digital, esp. HD streaming isn't still as prevalent outside NA and certain EU and asian countries as one might think primarily because of the ever-annoying bandwidth caps and also that the speeds aren't that conducive yet.

Personally, I bought a PS3 for a high price when I did because of BD as well as the future promised by Sony's first party.

Moving on, I'd assume that this time, the development of PS4's architecture would be more influenced by the complaints and necessities of first parties and a few key third parties.

 

[Rubenov]

So after all these years of eating humble pie dat douchebag attitude from Sony and Sony-affiliated enterprises is still intact I see.

 

[AngelSoldier]

So after all these years of eating humble pie dat douchebag attitude from Sony and Sony-affiliated enterprises is still intact I see.

ITT Microsoft has the douchebag attitudes.

What's on his disc better be whats on my disc ! (Even though Blu-ray has 50GB to work with)

If you release it on PSN first we won't let it on XBL !

Oh ! And you there ... Consumers ! Pay us to play your games in full !

 

[SmokedMeat]

You know what matters more than power? Selling more systems than the other guy, because at the end of the day, the company with the biggest lead in hardware sales will likely be the lead system of development. Publishers aren't going to optimize games for more powerful hardware if you're trailing behind in sales.