Image Image Image Image Image Image Image Image Image Image | December 12, 2017

Scroll to top



Are Cell Yields Really Doom and Gloom? |

PS3 Cell ProcessorChris sent in this link about the yields on Cell processors being really low. I’ve seen several other articles on the net about the same thing, but decided to hold off on posting about it until I got a second view.

NexTechNews – Poor production to plague the PS3/Bluray?

And here it is, from The Register. Yields of 10% to 20% aren’t very good. Sure, I’ll take that. It’ll make the Cell processor more expensive for Sony until they can increase the yields. Sure, I’ll believe that too. But the real problem for me is this: will it mean that Sony won’t be able to ship the PS3’s it promised? Not necessarily. The Register points out that yields of chips with 7 SPE cores operational are 10 to 20 percent. Cell chips with fewer than 7 SPE cores operational can still be used, even if they contain just one usable core. Personally I don’t know who uses these chips but my guess is that IBM would charge less for these chips and that customers would be found. But anyway – that’s getting off topic. We care about the chips with at least 7 operational SPE’s. With yields of 10 to 20 percent, will I still get my PS3? The Register did some napkin math to get the answer “yes”:

At 20 per cent yield, that’s 64 Cells per wafer, so Sony needs 93,7500 wafers to in 4.5 month to get enough Cells. That’s 20,833 wafer starts per month, which isn’t entirely out of order, given not only IBM but Sony will be producing Cells destined for PS3s.

So Sony might just squeek by. Of course, this calculation assumes that yields are 20% and we all know that they’re actually between 10 and 20. But who’s to say that yields won’t increase as time goes by? Actually, we know they will – that’s how these kinds of things work. The question is just how fast will yields increase?

So is it time to panic? No. Is it time to be concerned? Personally, I go by the motto “don’t sweat the little stuff”. So I’m not gonna worry about it. If it turns out to be a problem, then so be it. We’ll complain when the time comes.

Reg Hardware – IBM: Cell-like CPU yields 10-20 per cent

  • What’s gonna happen when you only get 6 or even 5 working cores instead of the promissed 7?
    If sony doesn’t deliver they might lower the quality in order to meet the demand, like having to lower the CPU frequency (cell is running at 2.8 GHz at the moment instead of the 3.2 as promissed. A year ago they said it would run at 3.8 GHz, all 8 cores. Now we got 7 @ 3.2, maybe 2.8)

  • You can’t give some gamers 5 cores and some gamers 7 cores. All the chips will have to be the same spec. Developers have been using and counting on 7 cores. The 10 to 20 figure is for Cell processors with 7 cores. What’ll happen is that the PS3 will be in shorter supply if Sony can’t get all the 7 core processors it needs. The point of this article was that that’s a big “if”.

    I’d like to see a link for your 2.8GHz and 3.8GHz numbers. I’ve never seen those. What I do know is that IBM was originally targetting 4.0GHz for the Cell, but when that didn’t work out, they throttled it down to 3.2GHz. At last year’s E3, Sony promised 3.2, not 3.8.

  • observer

    This isn’t news. It’s just the usual haters scouring for some bad news and trying to find a problem to spread fear about. Look for their actual source; it’s mostly nonsense.

  • Hasnt been said, that Sony started taping out the Cells some months ago already, stockpiling them, till the assembly starts?

    I cant quote that from anywhere, but I’m pretty sure bout that.

    There was NEVER talk about more than 3,2Ghz from Sony. IBM said things about more, but the PS3 ALWAYS were to get a 3.2Ghz Cell, nothing more, nothing less.

    The 2.8Ghz rumors bare no truth at all. These were estimations made up by TheInq (which is getting really specialized in bashing the PS3 these days) or other ubiquois sources…

  • observer

    Here’s the interview they’re getting that from.

    “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.”

    Logic Redundancy is where they only rely on seven out of the eight SPE’s. So he estimated 20 to 40 percent.

  • Matt

    A couple of things

    The cell chips with 8 SPEs working will not go into the PS3. I have read several articles saying that these chips are worth more and will be used for high end PC and Server applications.

    I have also read that if ONE of the SPEs stops working the PS3 is broken. It cannot run on 6 or less SPEs. Which if you think about it makes sense. Developers are designing games to run on 7.

    Sony and Microsoft both planned to use faster processing chips. 360 was spec’d at 3.5 and the PS3 was atleast that. This was also before the systems were announced so take that as you may.

    Whoever posted this. I think your numbers are off. In order to make 2 million units for launch, which have to be shipped from China, they have far less time than 4-5 months. They take atleast 8 weeks to cross the Pacific to get to the US. They need to have 600,000 to 1,000,000 units on cargo ships by September 1st to the US alone (also factor in retailer delievery). I also question the Cell percentages. 10%-20% is a huge difference in end results. To me it makes sense that they are closer to 10% but have had a few come off the line with 20%. I would bet it is closer to 12%.

    The bigger problem may be the Blu-Ray drive. The optical piece they need to make the laser is in extremely short supply. The 2 companies that make them are having problems supplying any Blu-Ray makers.

    I would expect a launch that is much worse than 360s. If you pre-ordered you should be fine but if you didn’t good luck. I strongly believe they will not ship 1 million units by the end of 2006. 360 had plenty of the main components (CPU, GPU, DVD drives, etc). It was a very small part or two that shut down production of the 360.

    I would strongly suggest selling your PS3 on EBay. You will probally be able to fetch $1000-$1500 for the “real” PS3. A good way to double/triple your money (HDTV fund?).

    This isn’t the end of the world. Sony will do everything possible to supply the world with the PS3. They just aren’t going to be able to make as many as they hoped and this could really hurt them if Nintendo and 360 have tremendous numbers of units to sell. Parents will buy something this X-mas to put under the tree.

  • observer

    I knew it. The original interview that spurred all this speculation is updated with a clarification. So basically none of this speculation was very relevant at all.

    From this link:
    “Tom Reeves, IBM’s VP of semiconductor and technology services, said he was not making any specific references to past or current Cell yields in an executive insight interview that ran last week. He was, instead, referring to large die yield challenges in general and the successful leverage provided by logic redundancy strategies. IBM does not release product specific yield information. This clarification was made on July 14, 2006.”

  • JordanR

    Assuming the yields are correct, in retrospect it would have actually been cheaper to go for two cells with 4 SPE’s each. True you would only have one more SPE in total, but you would also have an additional PPE which is nice for branch-heavy code.

    In the end, the yield problem may not be that terrible as the chips with 1 to 6 working SPE’s can be used for Sony TVs, Blu-Ray players, etc… They won’t throw away all that silicon.

  • inspectre

    Silence is a vacuum and the idiots are out in force again posting nonsense. Here’s what Kutaragi said last month.


    Q: Your planned shipping schedule is 6 million units worldwide by the end of March, 2007.

    Kutaragi: We are planning a monthly production rate of 1 million units. We have secured the parts required to reach this mark. This has been verified, so we should be set to go, barring any major oversights.

    Q: Does that include the “Cell” semiconductors?

    Kutaragi: No worries there. We began the manufacturing process last year (summer 2005), and now have plenty of them – enough to sell on the street, even. We’re hoping to provide servers using Cell’s on our side of the network in the very near future, so the more we have, the better. What was actually more troublesome was securing the generic parts required. As the economy is strong now, we had a hard time securing all the necessary parts to meet a 1 million unit / month quota – passive components, RAM, hard drives, circuit board materials, and even plating alloys.


    Of course NO ONE references this statement. The plaster a SINGLE SENTENCE taken out of context from an engineer at IBM who probably has no knowlege of actual chip yields. Man it’s hard to get decent information.