IBM backstabbing sony and neglected Apple ? (long read)

[rabidrabbit]

Of course, forgot that. Thanks.

 

[Rodéric]

Given an Article we've seen in the past I would say:

Toshiba's CELL idea : lots of mini cores (SPE whatever) only.
IBM's CELL idea : some Power cores only.
SONY's take : one Power core + lots of mini cores (SPE).

-XBox360 is powered by IBM's idea of what CELL should have been.
-PS3 is a mix of IBM's CELL idea & Toshiba's CELL idea, SONY's mixed.

 

[Carl B]

^^^Exactly, that's the 'History of Cell' action One posted. I'll try and find the link for the people who haven't read it yet.

 

[one]

Isn't this from some Apple fans circles? This theory seems to emerge often in such places and places like Ars Technica where PowerPC related things are discussed.

 

[Mefisutoferesu]

This seems kinda silly. You think Kutaragi was the one who went over the contract?

Nope, it was the fleet of Sony lawyers vs the fleet of IBM lawyers.

there is nothing they can do against it from a legal point of view since IBM put the lines ALL RELATED TECHNOLOGIES in their contract

Any giant greedy corporation lawyer would be ALL OVER a comment like that. IBM/Sony/Toshiba lawyers probably specificly stated what options each company has with various new endeavors to be dictated as they surface, meaning STI lawyers would have to conveine to discuss if IBM could do this, Sony does that, Toshiba does this and that. IF IBM is using this tech for MS/Nin projects then it is rights STI has specifically allowed, but I think that's silly. As far as I'm aware the projects were developed concurrently in sperate teams, so it's a tad difficult to finish one project and then apply the lessons to another project when neither finish 1st. As such, that means IBM was putting workforce that Sony payed for to other project, which if discovered means insta-lawsuit. "ALL RELATED TECHNOLOGIES" aside if your money is being used to do work for others, scope doesn't matter diddly and IBM is in for hell.

This to me seems like someone confusing the special sauce things MS asked for in the cores as IBM using STI work to improve the XeCPU. XeCPU core = Hollywood core (or is it broadway?) I'll consider it possible.

 

[Titanio]

Makes a lot of sense actually. But I don't understand why Sony would be upset, Cell is superior in just about every way to 360's CPU...

For IBM's wallet, a PPC-only chip would have been superior

 

[Carl B]

This seems kinda silly. You think Kutaragi was the one who went over the contract?

Nope, it was the fleet of Sony lawyers vs the fleet of IBM lawyers.



Any giant greedy corporation lawyer would be ALL OVER a comment like that. IBM/Sony/Toshiba lawyers probably specificly stated what options each company has with various new endeavors to be dictated as they surface, meaning STI lawyers would have to conveine to discuss if IBM could do this, Sony does that, Toshiba does this and that. IF IBM is using this tech for MS/Nin projects then it is rights STI has specifically allowed, but I think that's silly. As far as I'm aware the projects were developed concurrently in sperate teams, so it's a tad difficult to finish one project and then apply the lessons to another project when neither finish 1st. As such, that means IBM was putting workforce that Sony payed for to other project, which if discovered means insta-lawsuit. "ALL RELATED TECHNOLOGIES" aside if your money is being used to do work for others, scope doesn't matter diddly and IBM is in for hell.

This to me seems like someone confusing the special sauce things MS asked for in the cores as IBM using STI work to improve the XeCPU. XeCPU core = Hollywood core (or is it broadway?) I'll consider it possible.

Exactly, though the similarities are striking, some recent IBM article explicitly stated that the corporate firewalls on these console processors were pretty strong, so I would be surprised if the XeCPU and Cell PPC cores were anything other than similar conclusions based on soem common architecture both teams explored. Isn't the theory that the base core for the PPC was in fact some early IBM Power design from the early nineties or something that never saw mass production?

I swear referencing stuff would be so much easier if we created our own little in-house B3D Wikipedia or something.

EDIT: Here's the link to One's 'Origins of Cell' thread btw: http://www.beyond3d.com/forum/showthread.php?t=20563&highlight=Cell

 

[Shifty Geezer]

From those? I'd say pretty different. I mean the in-order processing alone makes it quite a shift architecturally.

But isn't that like the only major difference? The similarities between XeCPU and Cell PPe are circumstantial.

Consider it's 2001 and MS approach IBM. Cell isn't in development with Sony not working with IBM but developing EEE or E-Cubed themselves. MS say they want multicore, vector processing. They want small cores to fit more on. What would IBM's response be? Take a PPC's functional units, ALU, VMX, etc. Beef up the VMX, thin down the OOO to make a console chip (console's are traditionally in order, no?) that's smaller so they can fit 3 cores in. Have 'em share a cache and stick a few GPU access features in as requested, and there you go.

What influence would Cell development add over that approach? The similarities between XeCPU core and PPE are standards of PPC coupled with the requirements of an efficient closed-box platform - no need for OOO and smaller cores. The specialist features of Cell and XeCPU are very different. XeCPU has 128 register VMX on a fairly standard CPU structure. Cell has ring bus, local storage, yadayada, of which none of those unique features appear on XeCPU. So why think XeCPU borrows from Cell research?

 

[Carl B]

So why think XeCPU borrows from Cell research?

Well I don't necessarily. Remember I came into this thread saying that the article was more or less rubbish. Still it has always struck me that they came to those similar conclusions on the cores. Afterall Microsoft coming from the software background it does, and programmers in general not being too fond of this IOE stuff, I would have thought that they would have more naturally gone for an OOE setup. The ring-bus and the local storage is unique to the SPE-related portions of the architecture, so I'm not sure that's really relevent in terms of being different when what we're discussing is eery similiarities between the PPC cores.

Anyway it's neither here nor there, as we'll probably never know fully to what extent ideas may have crossed over.

 

[Shifty Geezer]

Well I don't necessarily. Remember I came into this thread saying that the article was more or less rubbish.

Didn't pay any attention. I just latched onto this one quote and starting arguing

I've talked myself in the past about how Sony must feel with the XeCPU cores being as similar as they are

Regardless of whther you agree with me or not I'll still continue to quote this one line

I've talked myself in the past about how Sony must feel with the XeCPU cores being as similar as they are

out of context and hold you to it even if you say it's not something you agree with, and I'll not except any valid response from you until you bow down to my superior reasoning and admit your inferiority.

 

[Mefisutoferesu]

I wonder is it possible that in order was a forced choice? Part of the reason OOE isn't really good for consoles (as far as I know) is that it's tends to eat up a lot of transistors (Intel can attest to that) and add a lot of heat (Intel will tell you so again), so I imagine that in order to get the kind of power at a realistic temperature they HAD to sacarfice OO? Or at least that was my reasoing for the similarities there, but... meh, others probably have better insight.

 

[Carl B]

LOL, well it's true - I have indeed talked myself in the past about that.

In fact I was pretty startled when I first saw the core details of the XeCPU. There's some other thread from back in the day around here where I'm basically wondering if they could possibly have been related in development, not via chance but via active human cross-talk, and indeed implying Sony must feel pissed/betrayed.

My problem with this article or whatever is that it seeks to 'blow up' something that has really always been there (at least for me) under the surface, but at the same time it's not really confirmation one way or the either, so it just kind of stirs the pot needlessly.

Anectodtely, it's around the same time I was wondering if the 'opening' up of the Cell architecture to third parties would mean that indeed Sony's console competitors could go to IBM for the next gen and seek variations on Cell to put in their own consoles, or whether or not there was some sort of non-compete built into the 'opening' of the architecture.

I'm only really interested in these things from the angle that I want to see how important 'protecting' the Cell architecture is to Sony in the console space, and how the business posturings of IBM, Sony, and Toshiba have left them in terms of Cell. Not anything that really matters for the short-term.

 

[rabidrabbit]

Can it be the other way round.
That Cell's PPC core is similar to XCPU, and that MS indirectly "funded" the Cell development
The xbox360 is releasing before PS3, but had the xCPU been in development before Cell?

 

[Carl B]

Well since Cell research began ~2000 and XeCPU research began in 2003 (or is it 2004?) I'm thinking that it probably wasn't the other way around.

Not saying it's *either* way really, but I think we know which way it's not for sure.

EDIT: You could be right on the OOE sacrifice simply being the logical choice due to die space Mefisutoferesu. I guess we'll need another interview or two from the MS side to be sure what exactly prompted that move on their part.

 

[Hardknock]

Does anybody know how much MS paid IBM for 360's CPU development?

Looks like IBM suckered both of them

 

[therealskywolf]

Xenon is IBM's "vision" for a GAming (Closed Box) processor. CEll is sony's vision for a multimedia processor.

I talked with a few people who work at AMD, and they said that while Cell is a very very interesting and powerfull piece of TEch it wouldn't be their1st option for a Gaming System processor.

Honestly, i think Cell usefullness as a Game processor is being blown out of proportion.

 

[PC-Engine]

Ok let me get this straight, somebody is claiming IBM used SONY/Toshiba CPU knowhow to design the PPE core? Please excuse me while I go laugh my as off.

Now that felt good. Now for the sad truth. SONY/Toshiba contributed NOTHING to PPE design. The only thing they contributed was some VAGUE mini cores IDEA with local stores. Even the SPEs were mostly designed by IBM. The only thing that came from SONY/Toshiba was some ideas from NEC's former supercomputer chip architect who now works for SONY and that was for the small cores ie "SPE/SPU/APU". Heck SONY/Toshiba didn't even know how to design these small cores other than "I want them to use local store.".

 

[Alpha_Spartan]

Makes a lot of sense actually. But I don't understand why Sony would be upset, Cell is superior in just about every way to 360's CPU...

Yeah sure...

 

[3roxor]

Xenon is IBM's "vision" for a GAming (Closed Box) processor. CEll is sony's vision for a multimedia processor.

I talked with a few people who work at AMD, and they said that while Cell is a very very interesting and powerfull piece of TEch it wouldn't be their1st option for a Gaming System processor.

Honestly, i think Cell usefullness as a Game processor is being blown out of proportion.

Then where would they want to use Cell for?

 

[Thegameman]

Doesn't Sony and MS own the Ip's of this chips in the first place?