Indeed, "a solid architecture that is sustainable" is the key.
But I think what's happening to the automotive industry and consequently to the chip suppliers who support the auto industry is this:
A slew of new technologies -- be it ADAS or connectivity -- are being introduced. The product cycle of a car would not change and yet the pace of technologies being innovated and introduced to cars is definitely picking up.
Safety, reliability, maintainability, and diagnostics are all needed for car applications. I'd think that a solid architecture that is sustainable would make a lot more sense than "reinventing the wheel" for every product cycle in every model of every car.
The quality requirements required a lot more investment. If your phone or computer stops working after the 1 year warranty, you throw it away and buy a new one. If your car stops working, you get it fixed.
Exactly. If that's the case, shouldn't the cost of automotive-quality chip development be shared by both chip suppliers and carmakers? Or, am I just dreaming here...
I would say that automotive suppliers are held to a much higher standard than suppliers to the computer and handheld industries. There were times that we would wrap dollar bills around the chips to stay in the business. The quality requirements required a lot more investment. If your phone or computer stops working after the 1 year warranty, you throw it away and buy a new one. If your car stops working, you get it fixed. There were times that cars were being parked in a field waiting for our chips. You can bet that the penalties cost more than the profit at that point.
@Junko:Look no further than Taiichi Ohno, a self-taught engineer who developed the just-in-time manufacturing system at Toyota -- and that was introduced in Toyota plants in mid 1950.
Speaking of Toyota, they got some local NYC press in an unexpected area. Their engineers consulted with a local soup kitchen on improving operations. The manager of the soup kitchen didn't see what auto industry engineers could do to make his operation more efficient, but they were able to analyze his traffic patterns and dramatically decrease the wait time to get in and get a meal.
Essentially, he was admitting people needing food ten at a time, so those waiting had to wait for ten seats to be clear before being admitted, seated, and fed. The Toyata folks changed it so that one of his people was dedicated to watching for seats to become free, and admitting people as seats became available for them. Wait time dropped from 90 minutes to ten minutes. This sort of process analysis to streamline operations and make things more efficient is what the Toyota engineers do, so it was no surprise they could apply thier skills to a soup kitchen as well as an auto plant. :-)
@AIPothoof:As you say, the market wouldn't support the cost of reengineering for down-sized parts.
I wasn't aware Checker made "stretch" variants, but it's not a surprise. There are custom coachmakers that will do stretch versions of whatever you like - I've seen a stretch Humvee limo tooling around NYC. But those are custom one-offs with prices to match, not an option you can order from the factory.
I think that's what killed so many of those early car companies: the only thing differentiating them was their body, everything else was interchangable.
Except that it wasn't. All of those folks were largely making all of their own stuff. It might not even have been interchangeable among their own models, let alone with anyone else's
I think what we saw was the inevitable consolidation that affects any industry. Back then, the automobile was a whole new thing, with a huge potential market. Lots of folks tossed their hat into the ring and started building cars. Costs were much lower, prices charged were higher and so were margins, and an auto company could make money making and selling a lot fewer cars.
As the market expanded, competition set in. Henry Ford revolutionized the industry with his innovations in assembly line manufacturing, which allowed him to make and sell the Model T at a far lower price than his competitors to a whole new class of buyers who coudln't have afforded previous vehicles.
Competition occurred. Some folks were successful and got larger. Some folks couldn't compete and folded or were bought by a bigger competitor. (General Motors was the result of comnining several smaller independent manufacturers into a larger entity.) Cars became the norm, not the exception, and most people had one.
In any industry, the eventual result is a few big players dominating the market and some smaller niche market players serving segments the big boys don't. That's pretty much what happened with automobiles in the US, with the Big Three dominating the market and others addressing specific niches.
The issues the Big Three faced came as the market and competition became global, and dominating the US market was no longer a guarantee of survival. Foreign automakers led by Japan entered the US marketplace with vehicles better than what Detriot was turning out and rapidly became major players. The Big Three are competitive again, but it took nearly going under and government action to force the changes that made them so, and their future is not guaranteed. I expect to see further consolidation in the global auto marketplace in coming years, and expect more brands to go under or get acquired. I don't see the Big Three as in immediate danger, but I don't see tham as invulnerable, either.
I suspect the big automotive chip suppliers (Freescale, TI, Infineon, etc) aren't under any more price pressure than, say, Apple's suppliers. I suspect they also like the longer, more pedictable automative lifecycle. (Yes, automative has its cycles, but they're a lot less severe -- on both up and down -- than semiconductor's).
From an embedded perspective, industries such as automotive and telecom infrastructure which are big AND have long product cycles are a gigantic plus. Spec in a x86 CPU or hot phone/tablet ARM chip and see if you can get it in a few years. On the other hand, my company has a board using the TI 320C6701 DSP which has been in production since 2000, and we still don't have any problems getting them (unlike buying a new Pentium III).
Their alternative would have been new downsized models of their own, but the market for purpose-built taxis wasn't big enough to support that.
Except that wouldn't match their model: they were't just custom coach builders, the frame was theirs too. And the combination was modular: if you needed another 2 benches, 6, more (I've seen "stretch" Checker limos with 10 but I don't know what the limit was) they could just drop another center section in the frame and body and weld them up. That meant certain sizes and strengths had to be adheared to. As you say, the market wouldn't support the cost of reengineering for down-sized parts.
"'Proprietary content" had a lot more to do with styling than components. Models were differentiated on what they looked like, more than by wat was under the hood.'
In a way, that's kind of my point: "badge engineering" isn't going to be enough. If the only thing unique about your vehicle (or other product) is the color of the paint, you're not going to hold market share. I think that's what killed so many of those early car companies: the only thing differentiating them was their body, everything else was interchangable. Since the bodies followed the fads, they all looked alike and the market was too small to sell enough to stay in business.
Blog Doing Math in FPGAs Tom Burke 20 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...