The Second Batch Info Thread

[Monk]

I'm going to answer OOO...

But isn't this fun?

Yes. Although there's riskj of confusing the newbies - wel, doesn't that pretty much describe a newbie? It's pleasant to have any kind of discussion where one can debate/consider possibilities without the fear of getting peoples backs up - there's been FAR too much of THAT, so this little threadette comes as a nice change

Actually, we don't know how TI are doing the testing. If they test at 720 and bin them in "can clock, can't clock" then we have guaranteed 720s and "can go up to somewhere between 600 and 720". So in that case, hypothetically of course, the 2nd batch should be overclockable by an unknown amount which varies from unit to unit, up to 720MHz. The 1st Batch will be able to overclock, and again this will vary greatly from unit to unit, anywhere from 600MHz upwards to an unknown ceiling.

I'm going to feel dirty after doing this, but it kinda supports my hypothetical situation - the more positive one I keep saying is POSSIBLE, even if I'm not confident of it enough to force a cancellation my first batch order:

It's worth knowing that we are yet to come across a chip which can't reach 720mhz.

Ptooft. I was right - I DO feel dirty. Now - where was I? Oh yes - that line supports my general proposal that there's no difference between the chips but marking. Imagine this hypothetical scenario.

1. TI makes some chips, and flogs them to run at "600MHz, at least".
2. TI tests a few chips and find they all run at 720MHz.
3. TI tests a few more chips, as they fly past production, and finds they ALSO all run at 720MHz. TI smile slyly.
4. TI tests a whole bunch more chips and find they ALL pass at 720MHz or higher. TI grin manically. They have A PLAN....

Times are hard. They can't raise the price of the existing product and say "Hey chaps, you can drive them faster" because they will lose customers who won't (or can't) afford to pay the higher price. They can't charge different customers drastically different prices because they will get called names in the playground (some variation, yes, but there ARE limits). So they hatch a plan.

5. Since they have found that ALL the SOCs work at 720MHz, they introduce a second product line - the same product, but "binned" as guaranteed to run at 720MHz "or faster". The binning process, in its entirety, consists of redirecting some of the SOCs produced to a different labelling machine, that stamps "720" instead of "600" on them. Voila! They can charge more for those items for the PROMISE/GUARANTEE that they will work at 720MHz. No more, no less, than a paperwork excuse to charge more.

Now - I can't say that this is exactly what IS happening, but (based on previous CPU manufacturer behaviour, as well as the twisted logic of Marketing) this is a realistic POSSIBLE scenario. Which would, in and of itself, mean that whether 2nd batch people got 600MHz or 720MHz SOCs would make no difference at all, functionally - they would both actually be getting the exact same silicon AND performance as first batchers. The only difference would be the shape of some ink, and the cost to OpenPandora.

I struggle to believe it, but it wouldn't be the first time something like that has actually happened.

I need to go and wash my hands now, after quoting Craig in a positive light. Yuck!

On the other hand, they may find they have a surplus of 720 chips and have to sell them rated at 600 because there's still a market for them, a la Intel.

All the 600's might be "720 chips" sold rated at 600 so that they can make money at both the "$high$" and "$low$" ends of the market, with only one manufacturing pipeline... but yes, the fact that Intel have already done this (more or less) in the past is what makes it seem possible, for me, that TI could be doing this on a wholesale basis with this SOC.

 

[BWoodson]

This is a common business practice used by lots of hardware companies. It's more cost effective to make 100,000 chips that can run at 800MHz and clock half of them down than to spend the R&D and extra manufacturing cost (the more you make at once the cheaper the cost) creating two distinct chips. Manufacturing always creates defects so they go through the chips and if ones are messed up so they can't get to the performance of the high end chip they'll sell it as a low end.

This is especially common in budget video cards with say half the pixel pipelines of the top end cards. Sometimes some of the pixel pipelines are actually faulty and sometimes not but the BIOS they flash the card with just shuts half of them off. In those cases you can sometimes get a budget card and flash it to unlock the full potential and effectively buy a top end card for a low end price. It's always a gamble though because you don't know before you buy whether that particular card is a budget card due to physical problems or if they just needed more supply.

 

[Dunny]

1. TI makes some chips, and flogs them to run at "600MHz, at least".
2. TI tests a few chips and find they all run at 720MHz.
3. TI tests a few more chips, as they fly past production, and finds they ALSO all run at 720MHz. TI smile slyly.
4. TI tests a whole bunch more chips and find they ALL pass at 720MHz or higher. TI grin manically. They have A PLAN....

You are an evil, evil man with a very dirty mind. I have absolutely no doubt that this is hypothetically true.

D.

 

[ximple]

It's been awhile since I've had my wafer fabrication class, but I seem to recall that the lower end chips were the chips located on the outer circumference of the wafer. This was due to the photo masking stage where the image that was projected onto the wafer would "blur" more on the ends thus having a higher probability of defect. I'm not sure how much has changed since then. It could be that TI would give the "center cut" so to say, the 720MHz blessing.

 

[logobogo]

I got reply from the pandora sales and was told that I'm somewhere around 1600 - 1700 in the order queue, which makes me believe I'm in the "second batch". So, any idea when the Pandora will be shipped to me?

This post placed some distracting numbers to my head, like 4+ months lead time on some parts (in a post back in January). What is the most up-to-date info?



I'm sorry if I overlooked something. I think it would be better to consolidate all of the information about second batch to this thread, if the information I missed was outside this thread.

 

[mali]

^ You are in the first batch.

 

[TrashyMG]

It's been awhile since I've had my wafer fabrication class, but I seem to recall that the lower end chips were the chips located on the outer circumference of the wafer. This was due to the photo masking stage where the image that was projected onto the wafer would "blur" more on the ends thus having a higher probability of defect. I'm not sure how much has changed since then. It could be that TI would give the "center cut" so to say, the 720MHz blessing.

Today's Fabrication is a lot better than it was a while ago, mask layering is a lot more accurate. Also with larger wafer sizes the photo layers are not done for the entire wafer any more, usually it's done for a small group of chips (1 to 4 usually, can be more) in a single reticle. This reticle is then projected repeatedly until the entire wafer is done. This eliminates that blurring effect you mentioned, however it can cause repeating fails if there photo mask is dirty or damaged.. but generally it's better than doing the entire wafer at once.

Generally defects of the levels of blurring masks would just not function, seeing how the circuit layers wouldn't even be working properly. The speed of the chips is usually determined by doing stress tests when they are in their completed form.

 

[logobogo]

^ You are in the first batch.

Hmm. Thank you. I'm still quite confused.

I keep reading "the first 1000" everywhere, but that isn't the first batch apparently. The first batch contains 4000 Pandoras? What kind of delay are we talking about after "the first 1000" Pandoras?

 

[WizardStan]

I keep reading "the first 1000" everywhere, but that isn't the first batch apparently. The first batch contains 4000 Pandoras? What kind of delay are we talking about after "the first 1000" Pandoras?

The original prototype cases were pretty good, but they weren't as awesome as OPT (OpenPandora Team) had hoped. The factory that is producing them promised that the final production cases would look much cleaner, that the prototypes were made with recycled plastic and lacked the finishing process that would make the case look smoother.
Rather than take the plastic company at their word (a wise move, given the communication problems they've had) they agreed to have 1000 production cases run off and shipped for assembly. If the cases turned out to be as good as promised, they would send word to the factory to do the remainder of them, which would take (by my estimates) an extra 2 weeks to be produced, finished, and finally land in the UK.
The thing now, however, is that MWeston has actually received some of these samples that have come out of the production process. EvilDragon has also mentioned that all parts for the cases have been produced and just need to be painted and shipped (which can't happen while flights to the UK are grounded anyway). However, what they haven't mentioned (either because they haven't thought to say, or nothing has changed) is whether MWeston looking at the production samples sent to him was enough to give the go ahead for all of them or remain with the first 1000, and if "all the parts" really means ALL or just all the parts needed for the first 1000.
Hopefully that clears things up a little.

 

[Dunny]

However, what they haven't mentioned (either because they haven't thought to say, or nothing has changed) is whether MWeston looking at the production samples sent to him was enough to give the go ahead for all of them or remain with the first 1000 [...]

It fits together great and actually passes Michaels QC.

Hope that helps somewhat.

D.

 

[WizardStan]

Right, but does that mean they told the factory to go for the full 10'000 (or at least 4'000) or are they still only expecting an initial 1'000?

edit: quote trim

 

[Dunny]

edit: quote trim

Your neatness and execution are to be admired, sir.

D.

 

[WizardStan]

Your neatness and execution are to be admired, sir.

And my mild OCD nature when it comes to forum posting etiquette?
(an etiquette of arbitrary rules I frequently change, I might add)

 

[Dunny]

Your neatness and execution are to be admired, sir.

And my mild OCD nature when it comes to forum posting etiquette?
(an etiquette of arbitrary rules I frequently change, I might add)

No, it was the elegance you employed in executing our little troll. I applaud you.

D.

 

[logobogo]

Now that we are so on track with the thread's topic, I must add my own magic here.


It seems these days one isn't allowed to say anything negative or s/he will be labeled as a troll. It's like in the real world when no politician has the balls to say no to "child pornography filters" being employed nationwide, because, of course, then s/he must be a pedophile.


And thanks. I'm now a bit more informed about the 1000-Pandoras thing. Thank you.

 

[Gruntfuggly]

^ You are in the first batch.

Delivery wise - isn't it actually the second batch now?

batch 1 - 1 .. 1000
batch 2 - 1001 .. 4000
batch 3 - 4000 .. ???

 

[gruso]

Nope, it's first batch all the way up to 4,000. From a manufacturing point of view it's been broken down into 1,000 + 3,000 for QC reasons, but in terms of orders & delivery it's still one big batch of 4,000.

 

[Gruntfuggly]

So as soon as the first ones start shipping, should we expect a constant supply until all 4000 are delivered? I sort of got the impression that the first 1000 (or maybe 700) would be completed and shipped first (with the resistor mod) but that there might be a gap until the rest of the first batch are fully assembled...

 

[logobogo]

So as soon as the first ones start shipping, should we expect a constant supply until all 4000 are delivered? I sort of got the impression that the first 1000 (or maybe 700) would be completed and shipped first (with the resistor mod) but that there might be a gap until the rest of the first batch are fully assembled...

Seconded. Official clarification on this is needed.

 

[Monk]

1. They are not going to make 1000 and then ship them. They aren't even going to make 700 (or 500, or 800) and then ship them. They are going, they have said, to ship on a daily basis.

2. The "First Batch" refers not to shipping per se, but to ordering. OpenPandora have (more or less, sortof, on a rolling basis) sold about 4000. That's the first batch of sales orders. They used that money to order the parts for the first batch. That's the first batch of purchase orders. Once they have been made, possibly after a short break to work on the software, they intend to sell a second batch. Then a third. the shipments do not correlate to the batch number on a 1:1 basis (for example, the first purchase order for cases is apparently 10,000 units).

3. There may be a time delay between the "first 1000" (cases) or the "first 700" (PCBs) or the (and so on) and the remainder of the 4000 units in the first batch. Whether this is minutes, hours, days, weeks or months is unknown at this stage. OpenPandora have had a go at predicting things, time-wise, in the past. It's better to wait and see than to ask for estimates. If they KNEW, and I mean really KNEW, that when the first 1000 Mass-Produced cases arrive they would be fine and they would give the go-ahead on day 1 to produce the rest (about 8500-9000 cases remain in their first case order) then they could tell the plastics factory that now. So they do not - can not - KNOW how long the wait will be, if any. The plastics company can theoreticlaly pump out cases fast enough that, if told soon enough, OpenPandora could receive new cases before they have finished assembling the first 1000/700/800/500.