new RPi model B planned soon

If the Model A is planned for March 2013, then RPF's banner headline will have declared the Pi as the $25 computer for a whole year by then.  Isn't there a law against such blatant long-term false advertising?

Morgaine Dinova wrote:

 

If the Model A is planned for March 2013, then RPF's banner headline will have declared the Pi as the $25 computer for a whole year by then.  Isn't there a law against such blatant long-term false advertising?

coder27 is being gamesome.  The link is to a 20 Feb 2012 comment where liz promises that "you'll definitely be able to buy Model As in March".  True, she doesn't specify the year.

I'm afraid we're not going to get anywhere in this discussion if you insist on reasoning that if a company has more fabrication capacity available to it through a buoyant fabrication sector then this implies that it will have longer wait times.  That's not logical, Sir.

When there are many parties able to provide a service then your wait times are inherently shorter because for any given amount of fabrication load in the sector, that load is spread over more fab providers with less queueing, and when you are tendering for service then naturally lead times are one important bargaining point over which there will be competition.  You can expect the average lead times to be lower in a buoyant sector with many providers, not longer.

Now if you want to argue that it's not that simple because different fab providers deal with different technologies then of course I would have to agree with you, it will inevitably reduce the set of providers which a given fabless manufacturer can utilize, but that wasn't what you claimed as far as I can discern.

Our most important point of disagreement isn't really that though.  It's that you seem to insist that it makes perfect sense for Broadcom to do nothing about assisting the non-stop manufacture of BCM2835 devices despite knowing since March the overall levels of interest, despite having an inside route into per-partner demand information through RPF, despite knowing that every chip manufactured in the current months will be sold because solid order queues exist for them, and despite being able to discuss the Foundation's 2013 plans with Eben Upton owing to RPF's intimate interdependency with Broadcom to ensure that there are no surprises ahead.

Instead of noting these special circumstances which create an effectively risk-free environment for BCM2835 manufacture for a certain number of months which Eben can undoubtedly quantify, you insist on treating the SoC orders as some anonymous corporate purchase request over which Broadcom has no knowledge nor control and which therefore results in an unacceptable manufacturing risk.  This just isn't so.  Pi is special.  Pi is as close to being a Broadcom product without officially being one as it gets.  The Pi alpha boards even had Broadcom's mark etched on them.  It's not some anonymous 3rd party that wants these Broadcom SoCs.  It's a family member, pretty much, and you can be confident that communications between the two parties are very close and very good.  Hindsight was not required by Broadcom, they have something far better, a direct channel to the right people and to the information they need to manufacture appropriate volumes safely.

Now because you haven't done so, I'll hypothesize one theoretical risk that might exist, hypothetically.  It might be considered possible that RS would reject any faster supply of SoCs if they were produced and offered to them by Broadcom outside of contract, willing to let their customers wither of old age in the wait queue as long as they don't have to invest more than some predefined $X in Pi manufacture per quarter.  Yes it's theoretically possible, and maybe you were implying that this would constitute a real risk for Broadcom, but is it likely?  (Your allusions to risk did not make the sources of risk clear at all.)

Well I really doubt that even the strongest detractor of RS would claim that there is any significant likelihood of that occurring.  I don't have any particular knowledge of RS, but even if they have some weaknesses, they are nevertheless one of those companies with a long pedigree that gets lumped into "Best of British", and the likelihood of them turning down SoCs to vanquish their backorder queue is remote, in my view, and AFAIK nobody has suggested that that is a real danger.  So from where does the risk and uncertainty that you've implied actually come?  It's really just FUD.  There can be no uncertainty nor doubt about any BCM2835 device manufactured not ending up on Pi boards and getting sold when this is done in concert with RPF as coordinator, and fearmongering about it is not helpful.

Lastly, I want to address your point about how come that Farnell and Roku have not been hit with major shortages as far as we know, but RS has been.  This is the only actual numerical evidence before us whereas everything else has been speculation, even if reasoned speculation, so this is important.  Well of course I don't know, and nor do you, we can only speculate again.  We have however been speculating about it for many months here, and a number of scenarios have been suggested in the past in various threads.

Deliberate under-ordering by RS despite full awareness that the length of its preorder queue would grow has of course been a leading candidate, but it was far from being the only one.  We also noted that because both manufacturers' websites collapsed when Pi was launched and remained in deep trouble for a lengthy period that spanned hours or days depending on who you asked and their location, the rate at which the two services picked up expressions of interest was more in the lap of the gods than balanced.  As a result, it's a very possible scenario that RS picked up many more registrations than Element 14 --- someone had to pick up more than the other after all.

Given an initial imbalance, we then discussed here various possible scenarios under which manufactured Pi boards were allocated to the two partners.  The favoured "equal numbers to both" scenario would have the consequence of  emptying the preorder queue of the partner with the shortest queue first, hence exacerbating an already bad situation.  In contrast, an allocation proportional to the length of each partner's preorder queue would empty them both in the same amount of time, but it was considered unlikely.  In other words, the partner that performed best in the chaos of the launch week and obtained the most registrations would suffer worst from the most likely allocation policy.  Well we know that the initial allocation policy from the first batch was equal numbers to each, so at least initially we know that this very thing must have occurred.  Do we know that it was RS that suffered most?  No, but it seems likely given that RS has always had the longer queue sizes.

So, things are complicated, and I wouldn't rush so fast to the conclusion that RS's problems are entirely a product of their failure with ordering.  It may have all started as a product of their comparative success with registrations, at least initially.  Once their own manufacturing got under way then of course they would have exacerbated their own predicament if they under-ordered or late-ordered, and it's never been denied that that is possible.  Despite RPF's chequered history with information release, nobody here automatically rejects information released by them (at least I don't), but only if there is additional cause to view something with suspicion, and I'm not aware of any such in this case.

Even if RS's problem was entirely a product of their own mis-ordering, it still doesn't eliminate the point that I was making earlier about Broadcom being a joint partner in all of this, and that the good business sense which they possess could have harnessed the extensive third party fabrication capacity that exists in the ARM ecosystem to help reduce backlogs further (not just for RS) without impacting on their more lucrative contracts, given that we've agreed that there is no shortage of external capacity.  I expect we're going to have to disagree on that because you don't want to embrace even tentatively the possibility that Broadcom's involvement with Pi is special and carries litte risk.  So be it.

Morgaine.

Almost everything in Ms Divonas post isn't right. I can't be bothered to quote all of it because its late, but most of her statements would need the use of a crystal ball to make a decision. And I've never got those dingers to work properly.

Face it Morgana, you are wrong here. I know you are finding that difficult to accept, but you have two people who have worked in the industry telling you stuff, and yet you still don't beleive them. Why are you so confident you are right, and the experts are wrong, in the face of so much contrary evidence? You are working from heresay and speculation, and have people giving you facts, and yet, still you thing you are right. That's mucho dingo dudette.

Actually, one point is worth adding - that of excessive capacity in the third party fabs industry. There is a lot of capacity for sure, but not SPARE capacity. If there was spare capacity, there wouldn't  be a long lead time - you could just get in there and use it now. So you can make a hell of a lot of chips, if you book ahead. Which means the alledged shortage now is due to an order not being big enough 3-4 months ago. Since fabless companies in this area are not in the game of making stuff in advance and storing it just in case, the ordering decision back them was underestimated. Which is understandable - who would have expected the demand to still be so high a year after launch. That's pretty unusual.

Billy:  Look up "argument from authority" and "logical fallacy".

Then come back to us with solid reasoning instead of your repeated "It's like this because I'm experienced and I say so."  It's not convincing in an engineering forum.  If you're truly experienced then it should be easy for you to provide a reasoned argument rather than your standard line.

mynameisJim and I have exchanged reasoned arguments without name-calling nor arguing from authority.  We may not agree but it's a logical exchange of views.  Try it, it will stand you in good stead.

Your last paragraph might even have qualified as logical if you hadn't incorporated another logical fallacy, a simple straw man.  "fabless companies in this area are not in the game of making stuff in advance and storing it just in case" is the straw man because it's not "just in case", it's to fulfil the solid and well-known back-orders for Pi held by the manufacturing partners.  I stated that plainly in the preceding post to which you were replying, so you're countering a premise that was never made.

Arguing bitterly over conjecture and idle speculation is surely a sign of a species with too much time on it's hands...

From a manufacturing point of view I have to concur that it makes little sense for a manufacturer (especially one with no actual manufacturing capability) to push an obsolete, low profit item to the head of a carefully planned queue. I would imagine that the main value of the 2835 / Pi thing to Broadcom is in terms of public relations, so possibly a beancounter in a Cambridge office knocked out a cost / benefit analysis a long time ago, while other grey suits got on the trumpet to their pet fabrication subcontractors - only to be given the option of being bent over a barrel. One has to look after one's important customers (and shareholders) first, so unless one can conjure manufacturing capability out of thin air and defy the laws of physics to shrink lead times, somebody is going to have to wait (especially if they didn't order enough in the first place). In fairness, it would have taken a crystal ball rather than a spreadsheet to predict the level of sustained demand that the Pi has seen.

Pretty shonky attitude there, Morgan.

Nothing I've said is wrong, or a straw man argument. And I reckon my experience actually working in this sector and actually knowing what I am talking about  trumps your inconsistent arguments. Don;t give me nonsense about reasoned argument  - you dont need to argue when presented with actual facts. You lost, admit it, or are you incacable of accepting you could be wrong about something. That's not reasoned argument, that being bloody minded.

For example, my statement "fabless companies in this area are not in the game of making stuff in advance and storing it just in case" is completely true and contains no straw shaped man. You in an earlier post stated Broadcom should make a load of chips up front without receiving  orders. I told you this is not the way they work. They make according to the backorders they are told about i.e by the people who actually know the numbers ie RS and Farnell.  They don;t make stuff that hasnt been ordered. They might make stuff up front if they are told by the manufacturer they expect to sell a particular number, but that owuld need to be a multi million chip order, not the small change orders fom the Raspberry Pi foundation. Some chip companies do make to fill warehouses, not companies like Broadcom, that's not their business. I dunno specifics, as I don;t work for any of them, but the only way Broadcom could know the actual order numbers required in the backlog is if RS or Farnell told them, and they way they do that is by ORDERING THEM. So they order enough for the backlog, plus the extra they expect to sell between receipt and the next order time. My guess is the RS completely failed to anticipate the demand would still be high and failed to order enough. Broadcom has exactly the same number of working crystal balls as anyone else, so without numbers from orders, they ain;'t gonna guess how many to make.

Jesus, its the middle of the night and I'm still arguing. People who cannot accept they have lost an argument do my blood pressure no good at all. Dunno why I bother. Must be the beers!

Billy Thornton wrote:

 

Pretty shonky attitude there, Morgan.

Every single post of yours is full of personal comments like that.  Try sticking to the subject and not attacking people but dealing exclusively with what the people say.  Don't refer to me, refer to what I am saying.  This will only cramp your style if you don't have a solid line of reasoning in the first place and therefore rely on abuse and ad hominem.  But if you do actually have something logical and reasoned to say then it will not cramp your posts at all.  Try it.

Billy Thornton wrote:

 

Nothing I've said is wrong, or a straw man argument.

I gave the precise details of your straw man argument, and it's not a matter of dispute because it was in your own words in black and white.  If you try to knock down an argument by adding a premise of your own that was not given and knocking down this new premise then that is the exact definition of a straw man argument.  It isn't a logical counter to what was said, but only a logical counter to the premise which you invented.  Don't do it.

Billy Thornton wrote:

 

For example, my statement "fabless companies in this area are not in the game of making stuff in advance and storing it just in case" is completely true and contains no straw shaped man.

I've gone over this before, but to help you along, I'll repeat what I already said.  The "just in case" was not what I suggested.  I very clearly suggested that they could work with Eben Upton and everyone else concerned to ensure that there was no risk by virtue of fabricating only enough parts to cover existing solid order backlogs.

If you put "just in case" in your counter then you are countering with a straw man argument.  It does not matter that the statement you are making is true when expressed without reference to a previous argument (in that case you would just be making a hypothesis yourself and debunking it).  When used in a counter to someone else's argument then it becomes a straw man because it does not deal with that person's argument but only with the premise you introduced yourself, and yet attributes the "success" in debunking it to the failure of the other person's argument when it is only a failure of your invented argument.  It exactly matches the fallacy type.  Don't do it.

Jonathan Garrish wrote:

 

From a manufacturing point of view I have to concur that it makes little sense for a manufacturer (especially one with no actual manufacturing capability) to push an obsolete, low profit item to the head of a carefully planned queue.

I agree with that, but once the game has started, we are where we are and it has to run its course.

When Eben was designing the Pi, issues of whether Broadcom would be interested in longevity of the BCM2835 should have been discussed.  Perhaps a chip in which they have more interest would have been a better choice.

It's a bit late to talk about that now though.  We are where we are, and Broadcom is lumbered with continued fabrication of an old SoC.  If they're unhappy about that then the best thing they could do is to get these blasted preorders out of the way by accellerating BCM2835 manufacture and at the same time work in concert with Eben to design the next version of Pi around a more modern SoC.

Morgaine Dinova wrote:

 

RPF has said clear as daylight that every Model A that could be produced would result in one less Model B being produced.

But nearly every Model A produced and sold is a Model B that would not be needed instead.

I think it is clear the reason As have not been made is that the distributors are not enthusiastic about them. 

Why make the even cheaper device and risk being left with waste stock of 9512s.

They have been very careful never even to gauge demand for the Model A so they can cling to the "too many people still ordering Bs" excuse.

Morgaine Dinova wrote:

 

RPF has said clear as daylight that every Model A that could be produced would result in one less Model B being produced.

But nearly every Model A produced and sold is a Model B that would not be needed instead.

I think it is clear the reason As have not been made is that the distributors are not enthusiastic about them. 

Why make the even cheaper device and risk being left with waste stock of 9512s.

They have been very careful never even to gauge demand for the Model A so they can cling to the "too many people still ordering Bs" excuse.

jopaji wrote:

 

But nearly every Model A produced and sold is a Model B that would not be needed instead.

 

I think it is clear the reason As have not been made is that the distributors are not enthusiastic about them. 

That's a pretty good point you make.  Even if one were to dispute the "nearly every" part, it's certainly the case that a proportion of Model A sales will replace Model B sales, so it's not plain sailing for the distributors.

Yes, I can see the disincentive for them.

[Addendum:  That said, the margin on Model A might be a bit higher, which would make them happier.  Perhaps we can estimate that.]