Bill Gates, then chairman and CEO of Microsoft Corp., once compared the rates of progress in the auto and computer industries, implying that carmakers were standing still. Soon after, somebody released this gem to the winds of the Internet:
If auto technology had progressed at the same rate as computer technology, a Rolls Royce would cost $20.00, get a million miles to the gallon and blow up once a year, killing everybody inside.
Rivalries aside, computer technology has infiltrated just about any industry you can think of, including anything that touches the automobile. As computers creep further in to automotive design, manufacturing, sales, repair and other aspects of the industry, innovators keep coming up with new applications for silicon. Those new uses often demand more horsepower of the binary variety.
Until recently, that demand wasn’t a problem. Moore’s Law states that the number of transistors on an integrated circuit (which represent computing power) doubles every 24 months, while cost stays the same.
This law, first proposed in 1965, has seemed more of a suggestion of late. Chipmakers were facing the end of the progress road they’ve been paving with silicon-based technology. Reducing the size of transistors means more will fit onto a silicon die, but the more transistors crammed on a die, the more heat they generate and the more electricity they leak.
Bernard Meyerson, Vice President of Strategic Alliances and Chief Technologist for the Systems and Technology Group at IBM, explained one of IBM’s woes. “When manufacturers tried to get more speed out of the chips, they made the chips so small that some of the layers in the chip, like the gate oxide (that we’ve replaced), were malfunctioning.”
For evidence of the industry’s difficulty in ramping up chip speeds, Meyerson looks to computer ads. “People rarely talk about gigahertz any more,” he said. “It’s extremely challenging to get higher-frequency parts out there.”
That’s why announcements from both IBM and Intel about their new chip technology have the semiconductor industry, and many of its clients, breathing a collective sigh of relief. While differing in materials and certain details, both plan to market chips using 45-nanometer transistors. Doug Cooper, Canada Country Manager for Intel of Canada Ltd., said his firm’s next generation core microarchitecture, code-named Penryn, will be a shrink of Intel’s Core 2 duo processor with a few extras. The smaller transistors switch faster and the new design consumes less electricity, so the chips deliver more performance per watt.
Effects of this advance on the auto industry won’t be felt for years to come, and then only indirectly. Computer hardware makers such as HP look forward to the new chips, but even they don’t yet have much to say, since the chips haven’t started shipping yet.
Meyerson notes that as industry players reached the limits of silicon technology, they moved towards more exotic, expensive technologies, and the cost kept them from becoming pervasive. “The new technology meets performance requirements using mainstream technology previously only achieved by more expensive technology,” said Meyerson.
Major technological advances don’t always coincide with new chip technologies. In Martin’s words, progress is more evolutionary than revolutionary. “The revolution was the introduction of computer technology. Everything else has been evolutionary.”
Advances in computing power can’t come quickly enough for many in the industry, though. Historically, newer technologies have driven down the prices of existing hardware, so the cost of many of the chips that continue to proliferate in cars today will fall.
Joe Barkai, Practice Director for Manufacturing Insights, an IDC company, separates the electronic components affecting automobile manufacturing into three bins. The infotainment group includes driver- and passenger- facing technology like navigation systems, upcoming night vision options, satellite radio and DVD entertainment systems.
Car control systems, like ABS and traction control, “don’t push the envelope as much” as the infotainment components, said Barkai. “Computers in cars today are relatively older architecture – either x86 chips or custom-made chips dedicated to a specific purpose, so that reduces cost.”
Cooper cited one concern: “Even though (the vehicle) carries a generator along with it, you can only generate so much power. You want to minimize the power you take away from the drivetrain.”
Barkai said the bugs that have plagued in-car systems as evidence that automakers need to simplify the increasingly complex electronics the put in their vehicles. He figures that automakers will achieve electronic reliability “if we can encapsulate higher functionality on a single chip.”
He also cited a lack of standard interfaces for electronics from one vehicle make to the next as an impediment to innovation. (Automakers haven’t yet agreed upon a standard.) Automobile controller area networks (CAN) are based on an older standard that worries Barkai. “When you start connecting newer devices, things get ugly,” he said.
Barkai added that the absence of camless engines might be due in part to their voracious appetite for computing power that traditional 24-volt car systems can’t handle. Even without camless engines, Barkai cites ongoing concerns among automakers about the continuing increase in electronic density.
The last category to be affected is design and manufacturing. Here, new chips play several roles to speed both processes and get products to market faster than ever, said Dr .Clemens Martin of the Faculties of Business and IT and Engineering and Applied Science at the University of Ontario Institute of Technology (UOIT) in Oshawa. Martin cites recent research that has produced better scheduling algorithms. These are currently limited to smaller shop floors that run up to ten machines. With more computing power, “models can be scaled to a more realistic problem size,” Martin said.
The smaller shops can benefit too. Martin envisions new forms of collaborative engineering, where smaller design companies build their modules and integrate them into a virtual model of a completely designed car. Currently, full 3D modeling is the domain of big manufacturers and tier 1 suppliers. “If it moves down the supply chain, design and innovation could speed up,” said Martin.
“Big manufacturers never had problem buying computing power they need,” said Martin. They could buy multiple servers to create more power to do processor-intensive work, so without higher-yield, lower-cost chips, smaller firms have difficulty buying state-of-the-art, expensive computing technology
That power can be especially handy when production goes awry. Most factories use what Martin calls a “plan fire and forget” approach to manufacturing. In such a system, “If something goes wrong – a machine breaks, people fall ill – you need to react to it,” said Martin. “We’re inflexible, so reaction is difficult.”
The trick for production control personnel is to quickly find their way around the problem without causing production standstills down a line of highly interlinked and complex systems. “When done manually, these decisions sometimes do more harm than good,” Martin said. “They fix an immediate problem but create a bigger problem a day later.”
“It’s hard for a human being to judge the ramifications of these systems.”
Simulation systems let production decision makers model the entire shop floor, every process, machine and person involved. Sophisticated what-if analysis provides a view of the consequences of a given decision on the line hours and even days later. In their quest for the most positive consequences, the more scenarios production managers can run, the better.
Fixing production problems may get easier as well. A Munich-based team of engineers with whom Martin is loosely affiliated is integrating 3D modeling into 3D glasses. In “augmented reality,” repair workers on a shop floor can put on their glasses in front of a given machine and see the schematics. To maintain or repair the unit, workers use the schematics in their glasses to diagnose and fix problems. This productivity booster also applies to
assemblers who put complex pieces together.
Such portable applications are the ones set to benefit most from more powerful, less power-hungry semiconductors. “That kind of a revolution, where you take silicon technology to a performance-price point where it had never been – that enables entirely new businesses and business domains,” said Meyerson.