• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

IBM VP comments on PS3 backwards compatibility

gofreak

GAF's Bob Woodward
There's an interesting, if sometimes confusing interview with Tom Reeves, VP of semiconductor and technology services at IBM, on ElectronicNews.com. It opens by talking generally about chip manufacturing issues going forward, but latterly turns toward Cell and Playstation.

Turn Down the Heat … Please

The most interesting comments are:

Reeves said:
Sony is very concerned about quality and backward compatibility. They want to get this right. They tested game after game after game. When there were about 40 Playstation 1 games that didn’t work properly, that didn’t pass their criteria for quality.

This would seem to bode well for the quality of BC on PS3 :)

Very confusingly, this comment followed:

Electronic News: So does that mean the current Playstation 2 systems have a Cell processor?
Reeves: No, they have a 440 Power processor. It’s a 130-nanometer, single-core ASIC chip. It’s the same technology as if you buy a Sony DVD or a Sony Bravia TV. Sony is replacing all the Mips design points with Power design points.

I emailed the article's author to clarify this, and he confirmed that this was the intended meaning - that Sony was now putting 440 Powers in PS2. I'm thinking there has to be a mixup somewhere, but I've asked a couple more questions and we'll see what happens. If anyone can figure out WTF he's talking about there in the meantime, do share..

The rest of the article touches on a couple of Cell issues - seems to indicate 20-40% yields for a chip "like Cell" when using logic redundancy. Also mentions that Sony does burn-in testing of all their chips when talking about reliability, and confirms plans to use 4/6/8-SPE chips elsewhere where possible.
 

demi

Member
Phew, I can sleep soundly knowing Bubsy 3D will work on my PS3, had me a little worried there
 

RaijinFY

Member
Electronic News: So does that mean the current Playstation 2 systems have a Cell processor?
Reeves: No, they have a 440 Power processor. It’s a 130-nanometer, single-core ASIC chip. It’s the same technology as if you buy a Sony DVD or a Sony Bravia TV. Sony is replacing all the Mips design points with Power design points.

Uh what?
 

gofreak

GAF's Bob Woodward
FWIW, the interviewer doesn't know anything more about that.

I'm wondering if the VP is assuming the chip is in PS2 now because Sony is "replacing all the Mips design points with Power design points.", but I can scarcely see how that would be possible with PS2. But maybe someone can enlighten.
 
not that I use PS1 bc but it's nice to know it's there. I never want to have to dig out old system if I feel like playing the original of some game.
 

Pellham

Banned
as long as the PS3 can play all PS2 games and at least my favorite PS1 games im not gonna care too much
 

Nightbringer

Don´t hit me for my bad english plase
gofreak said:
FWIW, the interviewer doesn't know anything more about that.

I'm wondering if the VP is assuming the chip is in PS2 now because Sony is "replacing all the Mips design points with Power design points.", but I can scarcely see how that would be possible with PS2. But maybe someone can enlighten.

I have a theory. The famous Cell for electronic home devices that Kutaragi talked few weeks ago is a PowerPC 440+2 SPE+DDR Interface is the CPU that Reeves is talking and it has pure retrocompatibility with PS2 because:

1. It is going to go inside the PS3 for PS2 BC.

2. Sony is making an antiWii system.
 

Animal

Banned
Nightbringer said:
I have a theory. The famous Cell for electronic home devices that Kutaragi talked few weeks ago is a PowerPC 440+2 SPE+DDR Interface is the CPU that Reeves is talking and it has pure retrocompatibility with PS2 because:

1. It is going to go inside the PS3 for PS2 BC.

2. Sony is making an antiWii system.

omg you just completely blew my mind.
 

DarienA

The black man everyone at Activision can agree on
demi said:
Phew, I can sleep soundly knowing Bubsy 3D will work on my PS3, had me a little worried there

This reminds me I need to harass the person who's gamespot quote made it on the cover of that game since I haven't spoke to him in a bit. Hey Chris! "...stunning...original...Bubsy 3D climbs back to the top...check it out!" --EGM

Never forget!
 

gofreak

GAF's Bob Woodward
Nightbringer said:
I have a theory. The famous Cell for electronic home devices that Kutaragi talked few weeks ago is a PowerPC 440+2 SPE+DDR Interface


IIRC, that wan't something Kutaragi talked about, it was his interviewer's own idea/theory..
 
gofreak said:
The rest of the article touches on a couple of Cell issues - seems to indicate 20-40% yields for a chip "like Cell" when using logic redundancy.

those yields would be incredibly expensive, the die size at 90nm is more than 200mm2 (somerthing like 220 if I recall), which is already expensive something on the magnitude of $100 I would guess. Throwing out 4 of 5 chips would cost sony $500 per cell. They could perhaps recover some costs if they were reusable (sold as 2/4/6 SPE cells)

And thats with redundancy, which they mentioned as not being currently implemented in cell.
 

gofreak

GAF's Bob Woodward
cleveridea said:
those yields would be incredibly expensive, the die size at 90nm is more than 200mm2 (somerthing like 220 if I recall), which is already expensive something on the magnitude of $100 I would guess. Throwing out 4 of 5 chips would cost sony $500 per cell. They could perhaps recover some costs if they were reusable (sold as 2/4/6 SPE cells)

And thats with redundancy, which they mentioned as not being currently implemented in cell.

"Self-healing" isn't implemented in Cell (i.e. if something goes bad with one of the cores on the chips subsequently, then unless that one was spare, you need a new chip). But obviously redundancy, and a fair bit of it, is there - with 1 SPE written off specifically for that in PS3's case, asides from any more general redundancy implemented across the chip. I think the latter is what he might have been referring to when talking about logic redundancy doubling the yield - and so those figures he gave may also be for 8-SPE Cells. (Or a chip "like" that, to be more precise).
 

davepoobond

you can't put a price on sparks
if true, why the hell would they change the processor? i wonder if those PS2 anti-Wii rumors are true now if they're making an effort to replace parts inside of the ps2 for seemingly no reason...
 

rubso

Banned
i don't know, but moving from MIPS arch processors, to IBM PowerPC-like Processor requires emulation software such as the Xbox 360 Emulator software to get PS2 Games working on Cell proc. because PS2 Games was compiled under MIPS arch ?
 

artist

Banned
why the f*** is he giving out yeild figures for CELL (or similar chips) anyway? It gives more credence to inquirer's story.
 
Interesting, do you think he's maybe confused equating to how Sony plans to implement BC on the PS3 with actual PS2 units? In other words, Sony will be emulating some parts of PS2 that they can't get to work in software right now on a power 440? Makes more sense than to replace the EE+GS chips that are currently used in PS2 production.
 

gofreak

GAF's Bob Woodward
I don't see why it would be easier to emulate a EE on a 440 Power than on a Cell..(and I do think it is probable that EE is emulated on Cell in PS3, and that the PS2 hardware included relates more to GS).

I don't know how likely it is, but I've seen the suggestion that the IOP in PS2 may have been replaced by the 440. Apparently the IOP changed in more recent revisions..?
 

flammie

Member
Reeves: Defects. It becomes a bigger problem the bigger the chip is. With chips that are one-by-one and silicon germanium, we can get yields of 95 percent. With a chip like the Cell processor, you’re lucky to get 10 or 20 percent. If you put logic redundancy on it, you can double that. It’s a great strategy, and I’m not sure anyone other than IBM is doing that with logic. Everybody does it with DRAM. There are always extra bits in there for memory. People have not yet moved to logic block redundancy, though.

Didn't IBM or Sony say a couple of months ago that the cell yield rate was improving faster than any other microprocessor beforehand? 20% (40% with redundancy used) yield, seems really, really low. How will Sony be able to produce enough processors for launch?
 
gofreak said:
I don't see why it would be easier to emulate a EE on a 440 Power than on a Cell..(and I do think it is probable that EE is emulated on Cell in PS3, and that the PS2 hardware included relates more to GS).

I don't know how likely it is, but I've seen the suggestion that the IOP in PS2 may have been replaced by the 440. Apparently the IOP changed in more recent revisions..?

True, I wasn't really thinking it through, but isn't a 440 a little excessive for just the I/OP?
 

gofreak

GAF's Bob Woodward
Possibly, though I don't really know enough about either to say for sure. I'm not sure how it could fit in, replacing the EE with it seems so unlikely.

flammie - such yields are not unusual for a new chip of that kind of complexity, I don't think. If you gathered data on the wafer output rate of Sony's plants and IBM's (those manufacturing Cell), you could estimate monthly output for a given yield (there's about 100 Cells on a wafer, IIRC). So for launch, you'd take that figure and multiply by the number of months of pre-launch production. However, I don't think Cell will be the bottleneck for PS3 manufacturing starting out, I think a more likely candidate would be something like blu-ray perhaps.
 

flammie

Member
cleveridea said:
those yields would be incredibly expensive, the die size at 90nm is more than 200mm2 (somerthing like 220 if I recall), which is already expensive something on the magnitude of $100 I would guess. Throwing out 4 of 5 chips would cost sony $500 per cell. They could perhaps recover some costs if they were reusable (sold as 2/4/6 SPE cells)

And thats with redundancy, which they mentioned as not being currently implemented in cell.

From a manufacturing standpoint, the cell doesn't cost so much. With a 300mm wafer costing maybe $3000 to produce, and with a 220mm die size, you can get maybe 270 Cells per wafer. At a 20% yield, they would only cost around $57 each. The major cost of any microprocessor is the R+D, which Sony and IBM have already poured billions into.

Edit: fixed wafer cost
 

WYWY

Member
Re: 440
Old news actually, and gofreak, you should know since you seem to frequent B3D.

To summarise, I have seen a few posts on B3D and 1 post on Ars forum(OT: which has a very large anti-Sony population) speaking on this. All recent PS2s no longer have the off-chip IOP. Everything truely sits on one chip only, We can safely assume the 440 now does the work of the MIPS core and the IOP,

Reason to do so? If it cuts costs they'll do it. Less external connections to extra sillicon, etc. If it replaces the 2 VUs as well - more power to them.

flammie said:
How will Sony be able to produce enough processors for launch?
I'm sure a lot of us are skeptical of the 6m number... it won't be the 1st piece of BS KK has dumped, and it probably won't be the last. If they achieve that, power to them.
 

Fafalada

Fafracer forever
WYWY said:
Old news actually
Well the identity of new CPU was a mystery to date - I'm actually not sure I even believe it is 440, but at the moment I have no way of verifying that anyway.

We can safely assume the 440 now does the work of the MIPS core and the IOP
440 isn't even remotely in the performance/feature range to emulate R5900 (let alone the entire EE as you suggested at the end).
Sony replaced IOP because it wasn't their IP to integrate - they have no reason to modify EE in that fashion because they own it.
 
Top Bottom