Why did Sony bother making the Cell for the PS3?

[Oninotsume]

I'm not the technical type, so I really don't get a lot of the reasons behind chip design and ram and all of that, but I get the general feeling that the PS3 is not really all that hot. Everyone keeps saying its lack of EDRAM is terrible, and that the overall system is really no different from the XBox 360.

So, my question is then why did Sony go through all the trouble to include cell in the PS3 if they could have gotten basically the same results as microsoft got with the 360s fairly standard chipset? Is the PS3 good at anything?

Sorry to be picky, but simple examples to coincide with the answers would be much appreciated. I'm not the technical type, I just want to know how it will effect the games.

Thanks

 

[pc999]

http://www.beyond3d.com/forum/showthread.php?t=25719

 

[liolio]

At the begining of the PS design, it seems that sony want to go with a multi cell design without a real GPU. the whole design is expensive because with this option you need hyper rapid ram and rambus technologie ($$$$) to keep the spe feed by the DMA engine of the cell.
The base of this architecture is streaming lot of datas.
I'm not hardware designer (lol by far) but it seems to me too that sony could have achieve a equivalent system in term of power without the expense of the lot of money spend in RD.
But the cell is still a powerfull design.
With the same "budget" as the ps3 i think a xbox360+ shall have prove to be more powerfull.

all piece of hardware in pS3 are more expensive than their xbox counterpart (difficult to say for the gpu rsx/xenos+edram sound the same).

still almost everybody agree that the ps3 has a edge in power against the xbox360.
it's ok for consumer but i think that in a sony point of view i would more bang for my bucks.

edit you say that the overall system is really no different from the XBox 360.
it's wrong the architecture are very different, perfomance wise we just can guess.
The XBOX360 is really a hype down product: you say fairly standart chipset
the xenos is a very innovative piece hardware
UMA
northbridge in gpu
smt
smart edram

Sony has really win the hype war with xbox1.5, the hardware of xbox360 is a very smart design.

 

[croc hunter2]

cell allows sony to use it for other things. it supposedly will be used for other things

as far as i can tell as long as the PPE is OK, cells can be used for all manner of electronics. digital tuners, dvd players, amps, anything that requires realtime decoding of digital info streams. a 1x8 cell in workstations, 1x7 in ps3, all the way down to 1x2 which could be used in a digital tuner.

all this ups the yield on each wafer. if sony went for a 1x8 in the ps3 they end up with a yield of say 10% but all the other chips are fubar. with a 1x7 it allows sony to use the 1x7s and 1x8s in the ps3 if need be. this might give them say a 30% yield. if they can use the other chips in other electronics they get more bang for their buck and more usable chips per wafer. therefore each chip costs less.


What plans they have for cell chips is anyones guess tho. so all the above maybe complete bollocks.

 

[Carl B]

The R&D was a lot at ~$500 million, but the billions went into fabs, something Sony would have done regardless of the design-path taken. People have to get over this hump in thinking that Cell is a multi-billion dollar project that exists in a void.

 

[London Geezer]

I'm not the technical type, so I really don't get a lot of the reasons behind chip design and ram and all of that, but I get the general feeling that the PS3 is not really all that hot. Everyone keeps saying its lack of EDRAM is terrible, and that the overall system is really no different from the XBox 360.

So, my question is then why did Sony go through all the trouble to include cell in the PS3 if they could have gotten basically the same results as microsoft got with the 360s fairly standard chipset? Is the PS3 good at anything?

Sorry to be picky, but simple examples to coincide with the answers would be much appreciated. I'm not the technical type, I just want to know how it will effect the games.

Thanks

Well, "why" Sony decided to make the Cell project is all inside Sony and Ken Kutaragi's brain really, but there are a few reasons, among which:
1) it offers double the floating point performance as the X360 CPU and many times the FLOP performance of top PC CPUs at this time. Whether that's any useful in practical terms is all to be seen.
2) that performance is not reflected in the cost (it does not cost double the XCPU or many times as much as top PC CPUs, much less in fact)
3) they can fab it themselves, which could save them a lot of money in the long run, but could also be very expensive. That's a risk Sony is taking and we'll have to see how it turns out.
4) they (and other companies like Toshiba and IBM and probably other licensees) can use it on lots of media-demanding boxes and workstations - this is all to be seen though, for all we know Cell might just be used in PS3 but all evidence leads me to believe the opposite.
5) Sony like to do their own thing, usually with Toshiba, and this time they got IBM involved too, which says a lot about the overall quality of the chip, whatever the haters say
6) this project will save them a lot of money in the future, as it is rumoured that the R&D used this time will be saved for PS4, which is rumoured to be heavily based on Cell, just a faster, bigger and better version of the architecture (lots more SPE's and PPE's for example). The R&D is already done and paid for, any improvement will cost a fraction of the original investment.
7) can't think of anything else.

 

[liolio]

cell allows sony to use it for other things. it supposedly will be used for other things

as far as i can tell as long as the PPE is OK, cells can be used for all manner of electronics. digital tuners, dvd players, amps, anything that requires realtime decoding of digital info streams. a 1x8 cell in workstations, 1x7 in ps3, all the way down to 1x2 which could be used in a digital tuner.

all this ups the yield on each wafer. if sony went for a 1x8 in the ps3 they end up with a yield of say 10% but all the other chips are fubar. with a 1x7 it allows sony to use the 1x7s and 1x8s in the ps3 if need be. this might give them say a 30% yield. if they can use the other chips in other electronics they get more bang for their buck and more usable chips per wafer. therefore each chip costs less.


What plans they have for cell chips is anyones guess tho. so all the above maybe complete bollocks.

you're right i guess sony wanted to get the cost scale down more quickly also by putting several cell in the ps3, but they 've changed their view on that point.

 

[Shifty Geezer]

Sorry to be picky, but simple examples to coincide with the answers would be much appreciated. I'm not the technical type, I just want to know how it will effect the games.

No offense, but if you're not the technical type, you wouldn't understand the answers. In simplest terms Cell is a special CPU designed with specific tradeoffs like all processors. The question of is it any good or not can only be answered after real work in the field. Any estimates up to now are pure speculation that can't factor in how devs do or do not manage to leverage Cell's stregnths. Any talk of PS3 is no better than XB360 (which has all custom chips too) is based on best-guesses, and may be right, may not. You can't really question Sony's choice of developing Cell without it ever actually having chance to prove if it works as well as hoped or not. And the reason Sony (and friends) developed Cell was because they believe it's strengths will be worth the development costs.

 

[_xxx_]

They will use the architecture in everything and anything that can connect to a network and will have instant 100% interoperability between set-top boxes, console, mobile phones, PDA's, servers, workstations or wherever they choose to use it. Of course, that will be stripped-down versions in devices requiring less processing power, but still based on the same architecture and with the same set of software tools. That's an investment in the future, if you want it that way. Unless they mess it up, that is

 

[liolio]

The R&D was a lot at ~$500 million, but the billions went into fabs, something Sony would have done regardless of the design-path taken. People have to get over this hump in thinking that Cell is a multi-billion dollar project that exists in a void.

don't take this a bias point of view what is this figure for MS and xbox360.
It could be an interesting apple to apple discussion ( if the subjet is already discussed, pleased
somebody sends me a link my english don't let me do efficient search in this forum)

 

[Carl B]

don't take this a bias point of view what is this figure for MS and xbox360.

I completely don't understand this sentence, but I'll assume you want me to think you're following comment isn't biased; don't worry, I don't.

It could be an interesting apple to apple discussion ( if the subjet is already discussed, pleased
somebody sends me a link my english don't let me do efficient search in this forum)

I would imagine less, no doubt. MS and IBM whipped this thing up pretty quick, and the impression I get is that the whole project was the model of efficiency, speed, and 'on-budget.' I don't have any news supporting the later, but normally you don't get the former without it. So I'd be shocked if it broke the $300 million barrier.

 

[liolio]

They will use the architecture in everything and anything that can connect to a network and will have instant 100% interoperability between set-top boxes, console, mobile phones, PDA's, servers, workstations or wherever they choose to use it. Of course, that will be stripped-down versions in devices requiring less processing power, but still based on the same architecture and with the same set of software tools. That's an investment in the future, if you want it that way. Unless they mess it up, that is

you seem to mess up hardware design and communication protocole between different devices.
most the format use by sony are IEEE compliant.
on the other side, cost scale down easily as crock hunter explain.

 

[cbarcus]

Bandwidth is key

A critical advantage of Cell is that it gives developers the opportunity to leverage most of its theoretical bandwidth. Looking at some of the performance tests for the processor confirms that it is a monster. The effect this will have on games will very likely be very noticable. I expect making full use of the processor will take time, but it appears that developers are already having great results.

 

[wco81]

I would imagine less, no doubt. MS and IBM whipped this thing up pretty quick, and the impression I get is that the whole project was the model of efficiency, speed, and 'on-budget.' I don't have any news supporting the later, but normally you don't get the former without it. So I'd be shocked if it broke the $300 million barrier.

Well there are those who believe IBM used the work on the Cell to provide MS with 3 modified PPEs on a die.

 

[Edge]

So, my question is then why did Sony go through all the trouble to include cell in the PS3 if they could have gotten basically the same results as microsoft got with the 360s fairly standard chipset? Is the PS3 good at anything?

That's your biggest misconception. CELL is a lot more powerful than the CPU in the Xbox 360. It has bandwidth numbers 2 to 3 times higher, and floating point numbers 2 to 3 times higher. That going to make a huge difference.

Also I would say CELL is more forward looking in achitectural design than the Xbox 360 CPU, simply because it forces developers to think differently when writing their programs. That's the more difficult programming that everyone is talking about, but lends itself to get the most out of the design and future designs. CELL will almost guaranteed be used in PS4.

 

[Carl B]

Well there are those who believe IBM used the work on the Cell to provide MS with 3 modified PPEs on a die.

Well, I used to be one of those in fact, and would still believe it. The thing though is that with the corporate firewalls that *should* have been in place between the development teams, it just seems outrageous that such a thing could have occured. I mean basically that would amount to MS getting a chip off of Sony's R&D dollars. Strange as it seems to have turned out, I just have to assume parrallel/coincidental development took place.

 

[liolio]

Well there are those who believe IBM used the work on the Cell to provide MS with 3 modified PPEs on a die.

i don't think it has take a lot of work for IBM to design a ppe core ore one of the three in xenon.
NO need to provide tips, Both are +/- in order evolution of ppc970 and the like.
IBM owns powerpc architecture.
you can probably say the same about the revolution cpu.
hence it seems that the xenon is more what IBM wanted the cell to be.

 

[Farid]

Search Function was created for these occassions

I'm not the technical type

Sorry to be picky, but simple examples to coincide with the answers would be much appreciated. I'm not the technical type, I just want to know how it will effect the games.

Not to sound offensive or incorrect, in any ways, but you're posting in a technical forum right now.
In other words you can't really expect from people to simplify things that are inherently complex, some might take the time to explain things by taking genral examples and making analogies, but it won't help you much if you're not interested in the technology in the first place.
What I'm trying to say is that there's plenty of threads that covers all of your questions already, and that it's up to you to catch up with this knowledge.

I'm not locking this thread, but I'll lock any other threads with this kind of premise, it's way too vague and already discussed in thousands of posts.

Edit: Actually I'm locking it seeing how there's already the almost same thread on the same page:

http://www.beyond3d.com/forum/showthread.php?t=25719