Weekly Edition
Return to the Press page
| |||||||||||||
Reflections on the hardware industry
Harald Welte
reflects
on the issue of insufficient FOSS support from the hardware industry.
"They all have some people inside their own organization, most often actual engineers or even engineering managers up to the very senior R&D managers who understand the FOSS model and the benefits that this would or at least could bring to their products and their organizations. They want to release the source, they want to push mainline, and they might even want to release the user manuals.
But inside the industry, nobody listens to what their own R&D department or event some external entity - even the very representatives of the operating system they use (Linux). The chip makers will only listen to one thing: Demand from their tier-1 customers. Whatever is in the spec of those who buy their components in millions of units will get implemented. Only those maybe biggest five board-makers are considered 'customer'. Everybody else is not."
(Thanks to Paul Wise).
(Log in to post comments)
Reflections on the hardware industry Posted Feb 5, 2009 21:31 UTC (Thu) by wmb@firmworks.com (guest, #56512) [Link]
I think that Harald's observations about what happens are mostly accurate, but I'd like to offer a different perspective on why companies tend to focus on their big customers and why it might in fact be in their best interest to continue doing so.
Harald wrote:
As a chip maker, your first and foremost concern should be to sell as many units as possible.
I'd argue that, as a business, your first concern is to make as much profit as possible. By and large, the companies that don't focus on that are no longer companies, at least not in the cutthroat merchant microprocessor market.
Profitability is related to unit sales, but profit also includes the cost of sales and support, plus the actual cost of goods sold.
Cost of sales is not linear with volume. A customer who takes 10M units probably only costs you, saleswise, 10x what a 10K unit customer costs. It might be even worse than that, considering that to get one actual 10K unit customer, you probably have to go through the sales cycle with a half dozen or more.
The other problem is predictability. Huge customers generate enough individual volume to let you plan ahead, so you can schedule production and secure supplies, raw material, and capital improvements necessary for volume production. In principle, you might be able to aggregate enough smaller customers to get to the same volume, but it's much much harder to get "real" estimates. Smaller customers tend to wildly overestimate their volume - if the didn't, nobody would pay any attention to them at all. And many of them never make it to production at all. (I've seen this effect from both sides...) Planning is absolutely critical for mass production. Getting it wrong severely affects your bottom line, and can easily kill you.
The predictability/planning thing affects the cost of goods sold. You only get the full effects of economy of scale when you get the planning right. That's especially true when you consider that you often have to spend money in advance to secure inventory and fab cycles, but you get paid down the road.
It's quite possible that you might not make any profit at all on sales to small customers. When that is the case, the only time that you (as a businessman) pursue small customers is when you happen to have excess inventory - with sunk costs - that the big customers won't take. And for the next build cycle, you try not to get into that situation again.
The behavior of big companies is definitely anti-competitive. A fundamental result of economic theory is that profits tend to zero in a competitive market. The goal of every big-company capitalist is to be a monopolist with plausible deniability. If you want to have some fun, use the "M" word in a public meeting with Intel or Microsoft people, then sit back and listen to the verbal dancing. (Of course, you might not be invited back to any more meetings...)
Reflections on the hardware industry Posted Feb 5, 2009 23:20 UTC (Thu) by jlokier (guest, #52227) [Link]
I'm sure you're right about the rational behaviour of big suppliers paying virtually all of their attention to big customers.
(Being a smaller customer, I know what it's like to receive zero attention from the big supplier, and the middle company who supplies me seems unable span the gap, resulting in chips which are quite capable, but grossly underutilised because the software for them is poor quality and closed to improvements.)
But I think Harald has a point in noticing that the big customers have many downstream smaller customers who want FOSS (and other forms of access) available for the platform, and the big suppliers have many internal suppliers who are ready and willing to provide FOSS (and other forms of access) to the platform, that all this would save everyone time and money, yet somehow, in the middle, the opportunity to connect these desires is lost.
In other words, I think Harald has a point that the rational behaviour of big suppliers and their big customers does not even produce what the big customer's customers what. Somehow the desires pass along the supply chain very slowly.
I suspect that's because most of the work done in this field has a certain "way it's done" which is really hard to change. Harald's observation about the companies in the middle is spot on:
"Those intermediaries (OEM/ODM) typically have very limited skill and understanding about anything related to Software, not even talking about FOSS. They know how to make many boards cheap. In fact, their skill typically is so low that all they can use for their products are so-called turnkey solutions: A full reference board design and complete software stack that they can copy+paste with only the most superficial modifications."
That "turnkey" approach in the intermediaries is probably getting in the way of more innovative and diverse potential downstream customers having their desires aggregated into big contracts at the 1M+ scale.
There's an opportunity in that, but I've no idea how to go about fulfilling it.
(Other observations about "intermediary" companies. 1. If there are significant bugs in the reference design, the small downstream customer is often stuck with them because the intermediary lacks the skill, resources or motivation to fix them, and the small downstream customer has limited access/permission to fix bugs despite often having the technical ability. They may even be prohibited by license from doing so. 2. There's an awful lot of wheel reinvention and working in closed groups. Fixes to kernels don't get mainlined; fixes to drivers remain as patches in obscure private trees, and even trivial board-specific or chip-specific drivers are provided as binary-only blobs.)
Reflections on the hardware industry Posted Feb 16, 2009 23:55 UTC (Mon) by tpo (subscriber, #25713) [Link]
> That "turnkey" approach in the intermediaries is probably getting in the > way of more innovative and diverse potential downstream customers having > their desires aggregated into big contracts at the 1M+ scale.
> > There's an opportunity in that, but I've no idea how to go about > fulfilling it.
The whole situation sounds similar to past ones where IBM was in the position to tell its customers what their best interest were and consequently the market routed around IBM. Same with Microsoft where the very powerful, new, innovative approach to route around the entity blocking the ecosystem through its market monopoly inertia was consequent open source SW.
Open Source HW where innovative end of the chain companies can expand and adapt and fix the HW designs to their need and by feeding them back lower their long-term costs seems like a logical answer to today's deaf chip makers. However whether the opened path will add enough advantage at low enough cost and how long it will take to reach comprehensive market capacity seems unscrutable. ?
Reflections on the hardware industry Posted Feb 9, 2009 17:21 UTC (Mon) by bfields (subscriber, #19510) [Link] Cost of sales is not linear with volume. Yes, of course, but I don't think that's the point. You certainly can't afford to sell directly to all your end customers. But you might still be able to afford (and benefit from) market research to determine what your customers' customers need.
Reflections on the hardware industry Posted Feb 6, 2009 1:45 UTC (Fri) by pabs (subscriber, #43278) [Link]
Another reaction to this post:
http://clunixchit.blogspot.com/2009/02/dont-read-this-lac...
Reflections on the hardware industry Posted Feb 6, 2009 17:20 UTC (Fri) by pboddie (subscriber, #50784) [Link] From the response: A more professional and long term approach would be wise. In other words: don't rock the boat! Meanwhile, and I'm not sure the author of the response really understood what kind of software was being discussed, the end-user ends up with unreliable device behaviour thanks to the pervasive "it works for us, ship it!" mentality at various levels in the hardware production chain, which is at the very least hinted at in the original article. The result is that once the precise combination of binaries keeping a device running can no longer be sustained, the end-user is supposed to throw the device away and buy a new one. It has been claimed that even the top-tier customers of the semiconductor industry can't get decent specifications under reasonable terms, where companies purchasing millions of units are too cowardly to make demands while presumably putting up substantial amounts of money. Some of these companies also urge a "don't rock the boat!" attitude as if their customers, dissatisfied with bad software, will somehow awaken a sleeping but bad-tempered giant who will take all the toys away. It's about time parts of the hardware business woke up and realised who is paying whose salaries, although I'm sure all the patent cross-licensing and other anti-competitive tricks will be used to squeeze out more forward-thinking competition (moves towards some openness from AMD and Intel notwithstanding) and push that day as far into the future as possible.
Reflections on the hardware industry Posted Feb 6, 2009 20:36 UTC (Fri) by Max.Hyre (subscriber, #1054) [Link] the pervasive "it works for us, ship it!" mentalityA (good) hardware engineer of my acquaintance was wont to say ``The question is not `Will it run?', but rather `Can it fail?' ''. Unfortunately, he had to say it a lot, due to encountering the aforesaid mentality.
Reflections on the hardware industry Posted Feb 7, 2009 20:35 UTC (Sat) by jlokier (guest, #52227) [Link]
The answer, for the moment, is perhaps to channel our purchasing and development energy to those companies like Intel who look like they "get it". Certainly for video display PCs, I've been recommending Intel instead of VIA for some time now, because Intel really mean it with good open source video drivers, and VIA have the rhetorick but don't get it at all, and this translates to real troublesome support issues.
Reflections on the hardware industry Posted Feb 6, 2009 6:55 UTC (Fri) by TRS-80 (subscriber, #1804) [Link] This gives a good reason for the FSF's lawsuit against Cisco - it'll get Cisco to put GPL compliance on its spec list.
Reflections on the hardware industry Posted Feb 12, 2009 17:21 UTC (Thu) by drag (subscriber, #31333) [Link]
There are two repsonses Cisco can take to the lawsuite
* CheckMark GPL compliance
Reflections on the hardware industry Posted Feb 14, 2009 5:31 UTC (Sat) by i3839 (guest, #31386) [Link]
The latter costs more money.
And the first can't be avoided for existing products anyway.
So their best option is to add internal procedures to make sure that they respect copyright.
|
|||||||||||||