[Panajev2001a]

This does not settle anything... have you ever heard of fabs that can produce limited but towrking demo units ?

In fact, rumours from within the Japanese electronics industry suggest that the first Cell chips could already be trickling off a special production line for testing purposes, with Toshiba - one of Sony's key partners in the development of the chips - currently working on the chip at its Oita production facility.

Special prodduction line for testing purposes != final production line for PlayStation 3's Cell implementation...

The two lines mainly dedicated to Cell are using 65 nm technology with 300 mm Wafers...

Their 65 nm manufacturing process is for e-DRAM, logic and logic PLUS e-DRAM as it would be needed in PlayStation 3's Cell implementation...

[nondescript]

Originally Posted by rabidrabbit

What is the logic behind that.
And I'm sure the Average Joe, who afterall is the biggest PS2 buying group, has not even a slightest clue what a 'Cell' is, or if it is in schedule or not.
I believe these early cell announcements are made mainly to keep shareholders happy.

Sorry, should have made that clear ... I was referring to previous statements. For example, not making it clear before that Cell would be on PS3 ... not stating the launch date, that kind of thing.

Obviously, if Cell is ready for tech demos and possibly SDKs by March 2004, there is a good chance PS3 will launch 2005. By refusing to disclose the launch date (and other relevant information), Sony can prevent people from "waiting" for the PS3. Who would buy the PSX in 2004 if the PS3 is coming out next year?

[Paul]

Guys.. we need to break into the Oita fab and see what they are up to. And maybe steal us a free sample.

*Breaks off into the night*

[nondescript]

Originally Posted by Panajev2001a

This does not settle anything... have you ever heard of fabs that can produce limited but towrking demo units ?

Special prodduction line for testing purposes != final production line for PlayStation 3's Cell implementation...

The two lines mainly dedicated to Cell are using 65 nm technology with 300 mm Wafers...

Their 65 nm manufacturing process is for e-DRAM, logic and logic PLUS e-DRAM as it would be needed in PlayStation 3's Cell implementation...

Ok, another thing I should have made clear... by "settles it", I meant that Cell is on schedule, PS3 is good for a 2005 launch if Sony so desires.

Of course mass-production cell will be on 65nm....these are the prototypes that Sony R&D and developers will use. I've read the Oita fab article several times now. By any chance do you know the litho technique Oita will be using? 65 nm using EUV, or the more traditional phase-shifting masks and OPC techniques. If there talking about shifting to 45 nm relatively quickly after that, I would assume EUV...

[Deepak]

From a thread from videogamereview.com

"For what it's worth, here's what I was told from my contacts at M$ and Sony, Atari (formerly Infogrames), and Vivendi Universal.

1) The recently distributed "specs" for PS3 are in extremely early form and are expected to change drasticly in the next 3 months (this is common). The PS2 specs went through 7 iterations in a year before solid specs/kits "core" programming even commenced on the games. IN fact they say that Sony has told them that they may postpone PS3's launch back to 2006 (American Release) because:

A) They can afford to wait one more year but mostly because:

B) They get to see what M$ is doing and can one-up/equate them.

According to Shane the M$ hard drive/no memory card thing stuck in Sony's craw and they don't want it to happen again.


2) Sony IS attempting to implement backwards compatability in PS3, but my friend Shane in Development at SCEA says that it depends on a number of factors (I'm assuming the ones Richard mentioned).

He also told me that it's almost a given that M$ Next Box will be backwards compatible for the simple reason that it makes sound business sense. "

*************

On the related note...how will PS3 play PS2 games? Through software emulation or by putting on EE+GS combo?

[Psychogenics]

Originally Posted by Deepak

From a thread from videogamereview.com

"For what it's worth, here's what I was told from my contacts at M$ and Sony, Atari (formerly Infogrames), and Vivendi Universal.

1) The recently distributed "specs" for PS3 are in extremely early form and are expected to change drasticly in the next 3 months (this is common). The PS2 specs went through 7 iterations in a year before solid specs/kits "core" programming even commenced on the games. IN fact they say that Sony has told them that they may postpone PS3's launch back to 2006 (American Release) because:

A) They can afford to wait one more year but mostly because:

B) They get to see what M$ is doing and can one-up/equate them.

According to Shane the M$ hard drive/no memory card thing stuck in Sony's craw and they don't want it to happen again.


2) Sony IS attempting to implement backwards compatability in PS3, but my friend Shane in Development at SCEA says that it depends on a number of factors (I'm assuming the ones Richard mentioned).

He also told me that it's almost a given that M$ Next Box will be backwards compatible for the simple reason that it makes sound business sense. "

*************

On the related note...how will PS3 play PS2 games? Through software emulation or by putting on EE+GS combo?

Would'nt software emulation make sense here since its cheaper.

[jvd]

I split this cause it was getting way to big. Here is a link to the first one
http://www.beyond3d.com/forum/viewto...er=asc&start=0

[notAFanB]

Would'nt software emulation make sense here since its cheaper.

I imagine that depends on the difficulty of etting 99%+ compatability. who knows maybe they just stick a cut down version of EE into the board.

[randycat99]

I don't understand what is such the big deal about alleged backwards compatibility in the PS3. Why is it so unbelievable that they would just drop in an EE (or squeeze it on-die somewhere)? (...and by extension of the EE, there will be PSX compatibility, as well) At 6.5 construction, it has got to be miniscule, almost a trivial thing to throw into the system.

[notAFanB]

Originally Posted by randycat99

I don't understand what is such the big deal about alleged backwards compatibility in the PS3. Why is it so unbelievable that they would just drop in an EE (or squeeze it on-die somewhere)? (...and by extension of the EE, there will be PSX compatibility, as well) At 6.5 construction, it has got to be miniscule, almost a trivial thing to throw into the system.

agreed although EE won't provide PSX compatability at alll, in line with the announcement that Sony will only support 2 system per generation.

[randycat99]

If it's no big deal to throw in an EE, then it will be really no big deal to throw in the I/O unit that goes with the EE (which would be the PSX). However, if Sony said otherwise, I guess that will be what will be...

[jvd]

Originally Posted by randycat99

If it's no big deal to throw in an EE, then it will be really no big deal to throw in the I/O unit that goes with the EE (which would be the PSX). However, if Sony said otherwise, I guess that will be what will be...

I think at this point its not really cost but fab space. Why bother making the ps1 tech at a very small micron when they can use that space to make more cell chip or ram or ps2 chips u know what i mean ?

[randycat99]

We're talking about such a small bit of die real estate, using that space for more Cell stuff would be inconsequential. I don't think people get this. You could probably weave it into waste die space where Cell could not expand into, it's so small.

[jvd]

Originally Posted by randycat99

We're talking about such a small bit of die real estate, using that space for more Cell stuff would be inconsequential. I don't think people get this. You could probably weave it into waste die space where Cell could not expand into, it's so small.

Its not like we are making a table and you can go get the scraps of wood and make a spice rack. They don't take the extra wafer parts and run more chips on them . If you can make 10 cells on a wafer and a 100 psones on the same wafer its one of the other . They both take the same fab space you just make more of the older one .

[Panajev2001a]

Originally Posted by notAFanB

agreed although EE won't provide PSX compatability at alll, in line with the announcement that Sony will only support 2 system per generation.

If you want to fully support PlayStation 2 you have to have a Hardware based or a software bases I/O CPU core which is basically the PSOne CPU+GTE+DMAC+MJPEG decoder, etc... you just have to... that chip is used in PlayStation 2 software and from that to play PSOne software is not that bad of a jump, quite the contrary if you ask me...

[randycat99]

You and I are talking about 2 different kinds of space on a die. You are talking about individual EE chips. I am talking about an EE integrated on the Cell chip itself in some potentially unused area of the die that cannot effectively be utilized by more Cell circuitry. It could happen either way, AFAIC. It will be such a small bit of silicon (or just regular sized at the original die spec, but at pennies of cost by 2005), it's pointless to worry about how in the world they could possibly "fit" it or "afford" it.

[Vince]

Originally Posted by randycat99

You and I are talking about 2 different kinds of space on a die. You are talking about individual EE chips. I am talking about an EE integrated on the Cell chip itself in some potentially unused area of the die that cannot effectively be utilized by more Cell circuitry. It could happen either way, AFAIC. It will be such a small bit of silicon (or just regular sized at the original die spec, but at pennies of cost by 2005), it's pointless to worry about how in the world they could possibly "fit" it or "afford" it.

An EE @ 65nm is roughly ~15mm^2; besides my munch is that the BE/Cell MPU will be cramped and hot enough. I'd think utilizing the preestablished EE+GS@90nm (Pana's) which will be already in mass production would be smarter; perhaps a further downgrade to 65nm eventually.

But your ultimatly correct in that it's pointless to worry for the reasons's you outlined.

[randycat99]

15? 16? Whatever it takes... :P I dunno. Is that a big or small number?

[jvd]

Originally Posted by randycat99

You and I are talking about 2 different kinds of space on a die. You are talking about individual EE chips. I am talking about an EE integrated on the Cell chip itself in some potentially unused area of the die that cannot effectively be utilized by more Cell circuitry. It could happen either way, AFAIC. It will be such a small bit of silicon (or just regular sized at the original die spec, but at pennies of cost by 2005), it's pointless to worry about how in the world they could possibly "fit" it or "afford" it.

wouldn't that cause more chips to fail thus making the cell chip more expensive ? (i will not even pretend to know everything about cpu manufacturing)

[Saem]

The problem with having the ee+gs with the one of the PS3 main PU, is that it'll likely introduce a critical path, which is ugly. Much better to have a dedicated seperate unit or software. Personally, I think the sofware solution will work fine since the PS3 will likely have lots of power to spare.

[Panajev2001a]

The PlayStation 3 will still need an I/O CPU ( the patent describes a separate I/O ASIC )

[london-boy]

I thought Sony explicitly said PS3 will not be able to play PS1 games, only PS2's... and that they would only support 2 platforms at any time, which means that the day Ps3 comes out, PS1 production and support will completely stop.... which isn't much of an issue. i dont think there'll be many people willing to play PS1 games in 2006... although someone might want to pick up their copies of, say FF7... nostalgia u see...

[randycat99]

That would be a shame then. Seems like a slam-dunk thing to pull-off if you are going to worry about any degree of backwards compatibility at all. Just keep embedding yesteryears model within last generation's model within the latest generation model, etc. As long as the die sizes (as far as transistor number) and complexity keep going up (especially at the rate Sony's consoles seem to be going), the embedded units just become more and more trivial as far as real estate and cost. FWIW, you also get the marketing benefit saying such and such new release console has a working library of many, many games extending back to 1997, for example. True, what real gamer would need to play those older games, but you cannot deny the marketing benefit of such a buzz feature (to the layperson). Nevertheless, the functionality is there. Plus some games transcend age, they are so good. It's a pleasure to come back to them every once in a while, if not to show somebody something you think was really, really cool from back in the day. Plus there's always the "newbie" who has never seen these old games before, and quite possibly could enjoy them if they can just see past the dated graphics. Take the "Greatest Hits Games" line, for example.

[Deepak]

One more CELL article...

Here

[zidane1strife]

If you can make 10 cells on a wafer and a 100 psones on the same wafer its one of the other

Well the cell should be about 1B+ trans. no? That's about 100x+ the EE... I dunno what's the trans count on the psx... but I believe cell could very well be several thousand times bigger.