What's the probability PS3 will start "PC era of consol

[Guden Oden]

What I mean by this truncated statement (subject line is short) is:

Sony's spending HUGE moneys to develop PS3.
They're also developing a scalable chip that will be used as a main processor.

Where am I going?

Consider that so far, each successive console generation has been a different hardware architecture to the one preceeding it. This is because of the explosive growth of the computer industry, and an architecture that was cost effective, perhaps even advanced a few years previous would be constrictive when it's time to design the successor. Therefore, out with the old, in with the new.

But! Is Sony aiming to change this pattern, ie, will PS4 be an up-scaled PS3 that will be able to run the same software without any kludgy solutions like including an IOP chip that is the main CPU of the previous generation etc?

It's said PS3 will include a high-level OpenGL-like graphics API, thus decoupling the graphics hardware from the game software. Cell is designed to be scaleable. Both factors point towards Sony wanting to make PS3 a "PC", in the sense it has the same basic hardware that just gets more powerful with time.

That's a cool thought, don't you think?

 

[Megadrive1988]

I know that alot of hardcore computer users would love to see a new AMIGA.

I think that a general purpose computer based on PS3 core technologies is a possibility.


The closest thing to a console PC would be the so-called 'Home Server' variant of PS3. like the rumored Xbox Next PC.

Sony & IBM want to compete with Intel/Windows. but how far they will actually go towards making a real product, other than PS3, for general home computing use, remains to be seen,

 

[GwymWeepa]

I was thinking the same thing the other day...I was wondering whether they would succeed in making an architecture that could scale up pretty much as much as they want, so say when ps4 comes out they have a few dozen "cells" (naturally on a much smaller die) as opposed to simply making a new architecture.

 

[London Geezer]

I was thinking the same thing the other day...I was wondering whether they would succeed in making an architecture that could scale up pretty much as much as they want, so say when ps4 comes out they have a few dozen "cells" (naturally on a much smaller die) as opposed to simply making a new architecture.

That's the idea. Not sure it will happen, or if Cell is the right architecture to achieve that goal, but i think IBM and Sony want exactly that. That's what they need if they wish to take over the PC business.

 

[cybamerc]

5 years is a long time. EE was also supposed to appear in new iterations.

 

[cthellis42]

5 years is a long time. EE was also supposed to appear in new iterations.

Well it DID, but not exciting ones. But hey, now we'll be getting it in our TV's! At least until it gets replaced by Cell or something.

At least architecture-wise, Cell has more likelihood to stick around for a while and be used in a broader set of capabilities, since that's supposed to be a function of its design, and it's supported by more companies in more ways than being "the PS3 chip."

 

[London Geezer]

What do EEs do in TVs? Genuine question.

 

[Guden Oden]

5 years is a long time. EE was also supposed to appear in new iterations.

Except the EE isn't a scalable architecture that's destined for multiple uses already from the outset. Not a comparable situation...

 

[cthellis42]

What do EEs do in TVs? Genuine question.

Erm... I think we had a few threads on this a while back. You can Google about for some quick info like this, though. Not much has been mentioned in other than marketing terms, though. To use more advanced graphical interfaces, encorporate more computing functionality when desired, and to eventually take advantage of the SoC design savings (when they can pump out more and the PS2 no longer needs it as much) I guess.

 

[London Geezer]

What do EEs do in TVs? Genuine question.

Erm... I think we had a few threads on this a while back. You can Google about for some quick info like this, though. Not much has been mentioned in other than marketing terms, though. To use more advanced graphical interfaces, encorporate more computing functionality when desired, and make more use of the SoC design savings.

Yeah and this is what it had to say about it...

Hardware-wise, the TV also mounts the Graphics Synthesizer (GS) graphics IC and the Emotion Engine (EE) microprocessor, from the PlayStation 2 (PS2) home-use game system. In the PSX the EE and GS are single-chipped using 90nm process technology, but the new flatscreen TV mounts the two chips individually. The design rule is 150 or 180nm.

We knew that... What does it actually DO?

 

[cthellis42]

Hence why I said "marketing terms." Just Google around some more or search out the old threads. Do you expect me to do ALL your work for you? Sheesh!

 

[London Geezer]

Hence why I said "marketing terms." Just Google around some more or search out the old threads. Do you expect me to do ALL your work for you? Sheesh!

You know who u're talking to, right. You do want a cooler place in Hell, the day you eventually get down here.

 

[cybamerc]

Except the EE isn't a scalable architecture that's destined for multiple uses already from the outset. Not a comparable situation...

It was planned for workstations and that never really took off. And they certainly had plans to keep developing the technology to the point where it would be used in a new PlayStation. So in many ways what Sony is promising now with Cell is similar to what they promised 6 years ago with EE.

 

[London Geezer]

Except the EE isn't a scalable architecture that's destined for multiple uses already from the outset. Not a comparable situation...

It was planned for workstations and that never really took off. And they certainly had plans to keep developing the technology to the point where it would be used in a new PlayStation. So in many ways what Sony is promising now with Cell is similar to what they promised 6 years ago with EE.

Yeah and if Cell turns out to be a dud, they'll change it again, until they get it right.
I don't recall Sony stating they would use the EE for workstations. They did play around with it with the GSCube's, but it was never anything more than big boys' toys/experiments.

 

[cthellis42]

I don't recall Sony stating they would use the EE for workstations.

Don't speak too quickly... The ever-popular Kutaragi in action.

I think one will find that the muscle, vision, and name behind IBM will do a lot more to validate the venture into other markets like workstations, though. And if the chips scale like they're supposed to (and ultimately provide the manufacturing savings they want) Sony and Toshiba will go a long way to integrating it into the consumer space.

 

[wco81]

I don't think media servers will have mass appeal. Having all your music and photos ripped and stored centrally is one thing but video is another matter.

People like PVRs but we won't see hudreds or thousands of movies stored in everyone's homes. MPAA won't give up the packaged media business any time soon. Plus video data is growing, not shrinking, despite the newer codecs.

So I really don't see a big demand for souped-up consoles with big hard disks. PVRs will become a popular appliance but they may become commodities that you get from your cable or satellite service for very little money. So no need to plunk down several hundred dollars over the price of a console.

That's not to say consoles can't have non-gaming functions. But what those might be remain to be seen.

 

[cthellis42]

Right, but it's not like that's the only option being talked about. People are simply tossing about "what else we may see it in" and whether Cell can successfully get branded outwards among a much bigger spectrum of devices.

 

[Entropy]

The Cell concept doesn't really come into its own until the number of processors is fairly high, the original IBM white paper dealt with 32-64 processors per chip.
The PS3 incarnation of the concept won't be close to that, presumably because it focuses so strongly (spends so much die area) on its vector APUs (128 architectural registers - I'd like to have a serious discussion of the limitations of extending x86 here some day).

If Sony/IBM aren't totally incompetent, and they shouldn't be, they will make sure that their tools will allow relatively painless scaling with the number of processors. That was one of the fundamental ideas behind Cell, after all - that you could more easily extend throughput with added processing elements rather than by making the individual processing element faster.

So it would surprise me a lot if the PS4 didn't reuse much of the PS3 design. That is not to say that it won't be polished, that they won't lift additional capabilities on chip to avoid off chip communication et cetera. New lithographic possibilities and limitations will always dictate a lot of implementation details.

 

[Megadrive1988]

I was thinking the same thing the other day...I was wondering whether they would succeed in making an architecture that could scale up pretty much as much as they want, so say when ps4 comes out they have a few dozen "cells" (naturally on a much smaller die) as opposed to simply making a new architecture.

I agree.

the PS3 was said to have 4 to 16 'Cells' and a few dozen APUs. though now that the GPU is probably not Cell-based, we're looking at more like 4 Cells, 8 at the most, but most likely 4. and still dozens of APUs.

PS4 should have dozens of 'Cell2's and hundreds of 'APU2's

 

[Megadrive1988]

Except the EE isn't a scalable architecture that's destined for multiple uses already from the outset. Not a comparable situation...

It was planned for workstations and that never really took off. And they certainly had plans to keep developing the technology to the point where it would be used in a new PlayStation. So in many ways what Sony is promising now with Cell is similar to what they promised 6 years ago with EE.

Yeah and if Cell turns out to be a dud, they'll change it again, until they get it right.
I don't recall Sony stating they would use the EE for workstations. They did play around with it with the GSCube's, but it was never anything more than big boys' toys/experiments.

in 1999, Sony had publicly announced grand plans for EEs to be used in workstations called Creative Workstations.

Phase I: workstations with 10x the power of TOOL/PS2 using multipul EEs (and GSs) running at higher clockspeed. planned for 2000

Phase II: workstations with 100x the power of TOOL/PS2 using multipule EE2s (and GS2s) with enhanced architecture over EE. EE2 would have 40,000 - 50,000 transistors. planned for 2002-2003.


Phase III: workstations with 1000x the power of TOOL/PS2 using new EE3 (and GS3s) with drastically changed architecture from EE. the EE3 would have 500 million transistors. planned for 2005-2006. Phase III silicon would also form the basis of PS3.

it was my belief that Phase III silicon: EE3 and GS3, were the Cell-based CPU (aka Broadband Engine) and one of the internal GPU solutions at Sony or the Toshiba GPU.


of these, only Phase I was put into prototype use in the form of the GSCube. well, GSCube does not exactly match the Phase I plan, but it does somewhat.


btw I think PS4 will realize all of the highest hopes that many people had for PS3. both inside and outside of Sony / STI.