Support NeoGAF

[Mark Gonzales]

http://www.reed-electronics.com/electronicnews/article/CA6350202.html?industryid=21365


Electronic News: Let’s look at design for manufacturability from a different standpoint. IBM has said it needs seven of the eight cores on the Cell processor to work for Sony’s Playstation. Will there be an aftermarket for chips with fewer operational cores?
Reeves: There are a lot of chips with six cores operational, and we’ve been thinking about whether we should really throw all of those away. We also have a separate part number for chips with all eight cores good. The stuff that’s going to be for medical imaging, aerospace and defense and data uses eight cores.

Electronic News: But might it be the less-expensive version of Playstation 3?

Reeves: It could, but I don’t think Sony has thought about offering that. That doesn’t mean there aren’t good uses for a chip with four SPEs [synergistic processing elements].

Electronic News: What’s the defining factor that makes some chips better than others?
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.

Electronic News: Do any of those cores ever go bad, so that you start out with seven and you wind up with six or five?

Reeves: There’s a reliability failure rate for all chip types. By definition, reliability failure is one point circuit that has failed. If it happens to be in an SPE, it will knock out one of the cores. We have electronic fuses now, rather than laser fuses, which you can only blow when you’re doing wafer tests. Electronic fuses you blow electrically. If you really want to be focused on reliability and up-time availability, you can design one of these chips to self-detect. You can ship it with eight cores working, blow one of them, and from a user perspective you would have self-healed it in the field.

Electronic News: But would it be as fast as the chip with eight cores?

Reeves: Yes, because the Playstation 3 only uses seven of them. You’d have a spare. That isn’t implemented in Cell, but it could be. We implemented that same strategy for IBM systems. If you take a logic hit on a chip, you don’t have any impact on performance because there is enough redundancy built in.

Electronic News: What happens if one of the cores blows on the Sony Playstation 3 if there are only seven to start with?

Reeves: It’s just like a reliability failure on your TV or DVD recorder. If it’s within warranty, you send it back. If it’s not, your game doesn’t work anymore. You’ll always have choices about how reliable you want to make a chip with burn-in. Most chips that go into the consumer marketplace on things such as camcorders or DVD players aren’t burned in. But you can add burn-in and improve reliability 5x to 10x. It’s extra cost. Certainly, a company like Sony adds that in.

Electronic News: How much extra cost?

Reeves: It’s variable. On DRAMs and SRAMs, it’s cents. On processors, because they’re so high-powered, it’s not trivial to power 100 or 1,000 at a time. With all the wattage, it can be dollars.

Electronic News: With the price Sony is going to charge, it can easily add that into the cost.
Reeves:

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.

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

 

[Juice]

 

[Liquid]

shortage.

 

[Mefisutoferesu]

http://www.neogaf.com/forum/showthread.php?t=109137

That title won't do any good for this thread. Please don't turn into another Sharukins.

 

[Speevy]

I don't know what that means, only who agrees and disagrees with it and why.

 

[davepoobond]

http://www.neogaf.com/forum/showthread.php?t=109137

That title won't do any good for this thread. Please don't turn into another Sharukins.

could it actually be him? i don't want to insinuate, but could it possible?

 

[Mefisutoferesu]

could it actually be him? i don't want to insinuate, but could it possible?

<shrug> I didn't mean to insinuate anything, it's just that he's been posting a lot of threads that quickly devolve because of the topic content. I guess I should have just asked to for him to please show a little restraint when posting. There's enough trouble on the forum as it is, having 2 sharukins is just... ... well, point is I'm just asking for a little restraint. Sorry, for any accidental insinuations.

 

[Trurl]

I don't know what that means, only who agrees and disagrees with it and why.

Ha ha, pretty much. To me it sounds like there might be nasty shortages but all this is way over my head. Hopefully a smart and trustworthy person can dumb it down for us.

 

[fse]

10% of the chips they make might work? Am I getting this right?

 

[Liquid]

Ha ha, pretty much. To me it sounds like there might be nasty shortages but all this is way over my head. Hopefully a smart and trustworthy person can dumb it down for us.

a good chip yeild is anything close to 90%. 95% is heaven. the higher the yeild the more chips you can get from a waffer. lets say a waffer could give you 100 chips at the absolute perfect. 90-95 chips is great and very cost effective. 10-40 chips is not only inefficient but a waste of fn money. BLEED SONY BLEED!!!:lol

 

[fse]

a good chip yeild is anything close to 90%. 95% is heaven. the higher the yeild the more chips you can get from a waffer. lets say a waffer could give you 100 chips at the absolute perfect. 90-95 chips is great and very cost effective. 10-40 chips is not only inefficient but a waste of fn money. BLEED SONY BLEED!!!:lol

shit, thats bad. how do they get better yeilds?

 

[GreekWolf]

So it's a possibility that the PS3 Cell might ship with 8 SPEs, as a fail safe in case one of them blows a tire?

 

[meltpotato]

a good chip yeild is anything close to 90%. 95% is heaven. the higher the yeild the more chips you can get from a waffer. lets say a waffer could give you 100 chips at the absolute perfect. 90-95 chips is great and very cost effective. 10-40 chips is not only inefficient but a waste of fn money. BLEED SONY BLEED!!!:lol

yeh uh... i really don't believe this news. 10-20% yeilds just seems too low to be true.

has there been any example of a chip that was in mass production like this putting out those yeilds for even a full fiscal quarter?

 

[Liquid]

time. the smaller the chips the higher the yields. this sounds like the chips are large. i dont get it. thats odd. i thought the chips would be small. sony shrank the chips in the ps2 by something insane like 90% IIRC or the PSone one of the 2 and the smaller the chips the more you get from a waffer and the more you profit.

 

[Mefisutoferesu]

Ha ha, pretty much. To me it sounds like there might be nasty shortages but all this is way over my head. Hopefully a smart and trustworthy person can dumb it down for us.

While I am neither smart nor trustworthy, it seems to me like he's simply explaning how redundancy can increase yields and seems to be using arbitary percentages for the purpose of an example. Defects in chips often lead to low yields, but by having redundant dram (memory) and logic (gates and so forth for calculations) you can increase your yields. For instance, with the CELL chip going in PS3 there are 7 SPEs (processing cores) within the chip, but only 8 are used, because of this if a defect on one of the SPEs occurs they can still use the chip, since only 7 working SPEs are needed. If 8 were needed and 1 SPE had a defect, they'd have to through it out.

 

[Liquid]

i think sony had the same problem with the ps2 in the beginning though so its not really insanely unbelievable.

 

[Mefisutoferesu]

i think sony had the same problem with the ps2 in the beginning though so its not really insanely unbelievable.

No, not even close. The PS2 had intial problems because Sony was producing them on a new process (I believe 130nm, but my memory's hardly the greatest). They were producing a large chip (for the time) on a brandnew process that still didn't have the bugs worked out. With CELL Sony is using a 90nm process, which is EXTREMELY mature and one they're well adjusted to. That's not to say yields will be great, but it's hardly the same situation.

 

[Liquid]

oh. well either way yields were effected then and something is effecting em now. shortage.

 

[Rhindle]

i think sony had the same problem with the ps2 in the beginning though so its not really insanely unbelievable.

20-40% yields on a CPU for a mass-market product actually is really insanely unbelievable.

Then again, not much about this interview makes much sense.

 

[firefoxsux]

Reeves: There are a lot of chips with six cores operational, and we’ve been thinking about whether we should really throw all of those away.

oops missed by one!

where can we write an email to ibm to request to collect thrown away cell? :lol

 

[Liquid]

20-40% yields on a CPU for a mass-market product actually is really insanely unbelievable.

Then again, not much about this interview makes much sense.

has anything like CELL been produced in the past though? i mean i'm not versed in this at all i just know this stuff because i just read about it in 360 uncloaked but is CELL like an 8 processor chip? 360 having what 3? and ps3 having 8? or does it not work like that?

 

[sharukins]

and PS3 production has not even started yet compared to Wii already started

 

[Count Chocula]

one-by-one mean 1mm^2 size of the die. That's when they're getting 95% yield.

Cell is 235mm^2

 

[flammie]

If you want to go through some rough numbers:

(Note, these are all estimates)

1. Yields average 20% from March until October, so 8 months.
2. Sony allocates 10000 wafers a month at $3000 a wafer.
3. Sony and IBM are using 300mm wafer diameters.
4. The Cell processor is 220mm^2.

To roughly caculate die per wafer:
wafer area / die area - edge of wafer partial die
(wafer diameter/2)^2*pi/(die area) - (wafer diameter)*pi/sqrt(2*die area)
So the total number of Cell's per wafer is (300/2)^2*pi/220-300*pi/sqrt(2*220)=276.

20% of them are good, so you get 55 functional Cell per wafer.

55*10000*8=4.4 million cell chips for launch
At $3000 a wafer, you get $240 million dollars spent total to produce those 4.4 million cells.

If Sony only needs 2.2 million cells, they could run 5000 wafers a month, and only spend $120 million on wafers for Cell before launch.

 

[sharukins]

could it actually be him? i don't want to insinuate, but could it possible?

:lol *devil*

 

[Speevy]

:lol :lol :lol

 

[Adam Blade]

:lol :lol :lol @ new thread title...

 

[davepoobond]

:lol *devil*

oh god. he's a doppelganger :lol

 

[Chris Remo]

So it's a possibility that the PS3 Cell might ship with 8 SPEs, as a fail safe in case one of them blows a tire?

As far as I know, that's not a possibility, that's the plan. They're manufacturing 8-SPE Cells, but the console will only ever use 7 of them. Some PS3s will have 8 functional SPEs, some 7, but it makes no difference because only 7 will be used.

 

[kinoki]

Didn't Kutaragi say something about sticking two 6 SPEs in a server unit?

Also, I'd imagine that the 4 SPE units will come to good use in HD-TVs.

 

[gofreak]

As far as I know, that's not a possibility, that's the plan. They're manufacturing 8-SPE Cells, but the console will only ever use 7 of them. Some PS3s will have 8 functional SPEs, some 7, but it makes no difference because only 7 will be used.

I think he says that "self-healing" is not implemented in Cell, so you'd only be able to keep going if the right SPE became faulty (i.e. the one not explicitly in use).

Anyway, this is all old. As I said in the original thread, he's suggesting 20-40% yields with logic redundancy - and that's possibly for 8-SPE chips if there's general logic redundancy across the chip. Or perhaps it's a reference to chips with 1 SPE redundant (i.e. 7). Low yields are not unusual for a relatively new chip. If you want a point of comparison, Chartered reaching over 50% yields on AMD chips back in March was almost a cause for celebration.

Again, could be wrong, but I don't think Cell will be the bottleneck in PS3 production. BTW, does anyone knows the wafer output of Nagasaki and East Fishkill for Cell? Chartered alone reportedly will be churning out 5000 wafers/month for Sony.

 

[flammie]

I think he says that "self-healing" is not implemented in Cell, so you'd only be able to keep going if the right SPE became faulty (i.e. the one not explicitly in use).

Anyway, this is all old. As I said in the original thread, he's suggesting 20-40% yields with logic redundancy - and that's possibly for 8-SPE chips if there's general logic redundancy across the chip. Or perhaps it's a reference to chips with 1 SPE redundant (i.e. 7). Low yields are not unusual for a relatively new chip. If you want a point of comparison, Chartered reaching over 50% yields on AMD chips back in March was almost a cause for celebration.

Chartered was selected as AMD's foundary partner back in August/September, and achieved 50% yields in under 7 months. And its not like x86 processors are simple to make. How long have Sony and IBM been working on Cell? 3-4 years now? 10-20% yields is pretty bad, especially with a mature 90nm process.

It seems from the article, that he's suggesting that IBM is the only major company to use logic redundancy, and that the 7/8 SPE is a different kind of redundancy than the one he is talking about when mentioning the 10-20% to 20-40% yields.

 

[gofreak]

Chartered was selected as AMD's foundary partner back in August/September, and achieved 50% yields in under 7 months. And its not like x86
processors are simple to make.

It's not like Cell is simple to make either! It's north of 200m transistors, which is well higher than the core duos for example.

How long have Sony and IBM been working on Cell? 3-4 years now?

They have pretty obviously not been manufacturing Cell and ramping yields for 3-4 years..

It seems from the article, that he's suggesting that IBM is the only major company to use logic redundancy, and that the 7/8 SPE is a different kind of redundancy than the one he is talking about when mentioning the 10-20% to 20-40% yields.

Correct me if I'm wrong - and I may be, totally - but wouldn't logic redundancy be something built-in at the design phase of the chip?

 

[flammie]

Correct me if I'm wrong - and I may be, totally - but wouldn't logic redundancy be something built-in at the design phase of the chip?

Yes, redundancy can only be implemented during the design phase. I think it is unlikely that Sony included logic redundancy, since the 7 SPE thing is already a form of redundancy. Also, logic redundancy could drastically increase the number of transistors per die and thereby increase die size, which is a huge cost issue. I mean, which parts would you make redundant copies of? In most chips, a large portion of the logic area is necessary for full functionality so you would need to duplicate a lot of logic. Sony probably was expecting high enough yields on Cell that most forms of logic redundancy would have be detrimental to cost.

 

[Kabouter]

It's not like Cell is simple to make either! It's north of 200m transistors, which is well higher than the core duos for example.

Really?
Because many DC AMD chips have ~233 million transistors.
I'd imagine Core 2 Duo's would have a similar amount.
And Kentsfield, well that'd just be insane I'd imagine.

Google says the Smithfield Pentium-D had 230M transistors as well.

 

[gofreak]

Yes, redundancy can only be implemented during the design phase. I think it is unlikely that Sony included logic redundancy, since the 7 SPE thing is already a form of redundancy.

I'd say if they implemented logic redundancy, at the time that decision was made Sony probably didn't know whether they'd need to be using a SPE for redundancy or not. I thought Reeve's comment might have hinted at them using it in Cell, but then he was talking about "chips like Cell", so who knows.

Really?
Because many DC AMD chips have ~233 million transistors.
I'd imagine Core 2 Duo's would have a similar amount.

Oops, I was looking at a figure for core duos. But Cell is in that range also, depending on what revision you're looking at (234m transistors was the first revision's figure).

 

[Suburban Cowboy]

So PS3 will be just like PS2, but not as dependable? 1337

 

[kaching]

I don't really see Sony or IBM being caught off guard by the initial yields here. Sony announced the PS3 would use a 7-SPE configuration of Cell more than a year ago now - they've known what they'd be dealing with for initial yields.

 

[Nerevar]

He's saying with no logic redudancy and needing all 8 cores, their yields would be roughly 10-20% per wafer. But by using logic redudancy and only needing 7 cores, they can increase their yields for PS3. This also ignores that they (IBM) are not throwing away cell chips with 6 or fewer SPEs functional (he mentions as low as 4), they're going to find other uses for them. In other words, nothing to see here, move along.

 

[<nu>faust]

i'm too lazy to read and try to understand what really goin' on here.... can some tech savy person summarize this article in a way that faithful gaffers can understand (ie, "sony am doomed" or "sony is teh roxxorzz") please ?

 

[J-Rzez]

Didn't IBM already say the Yields were rediculously high as is? I remember reading that a while ago... think B3D and such had a bunch of articles on that, and that was a while ago (too lazy to sift through the threads)... and that they were had ways in the works to even further improve that? So, not buying this guy... naw Sony ain't doomed, sorry to it's "fans"...

 

[evil solrac v3.0]

they said the exact same shit when the PS2 came out.

 

[meltpotato]

i'm too lazy to read and try to understand what really goin' on here.... can some tech savy person summarize this article in a way that faithful gaffers can understand (ie, "sony am doomed" or "sony is teh roxxorzz") please ?

Sony production am doooomed!~ but its okay because sony content and support r0xx0rz. and sony production am likely not doomed anyway...

that's pretty much all sides of the argument. i hope that helps.

Edit: Sony can do no wrong, Sony can do no right