
The effects of the global economic downturn are being felt hard by Japan's semiconductor manufacturers. For most of the last decade, many Japanese semiconductor manufacturers have been committed to producing systems-on-chips (SoC). Unfortunately, SoCs have not delivered the degree of success that had been hoped for.
A workshop, therefore, was held recently to help SoC developers move forward by putting their SoC knowledge and expertise to use. The Large Scale Integration (LSI) & System Workshop, held by the Electronics Society Technical Committee on Integrated Circuits & Devices of the Institute of Electronics, Information & Communication Engineers (IEICE), is a revamped version of the System LSI Workshop held throughout last year.
The System LSI Workshop of 2008 primarily featured SoC hardware developers as speakers, and was a meeting pretty much by and for SoC experts. The LSI & System Workshop, however - which began this year - features very few integrated circuit (IC) hardware developers as speakers. Instead, engineers are invited from fields unfamiliar to hardware developers, and include software developers, system developers, and system users.
The first presentation was given by a software expert, who said, "Performance is not really the priority item for software developers any more. We are more concerned with things like how easy it is to develop and maintain. That's proven by the fact that languages like Java and Ruby are being used." Hardware designers tend to emphasize performance, so to be told that performance was not the overriding priority came as quite a big shock.
The same software expert added: "There is increased interest in developing single instruction multiple data (SIMD) multiprocessor core ICs recently, and that's making it harder for software designers. There are many cases where we write a program for this type of chip and it doesn't run the first time. Even worse, hardware designers are interested in simultaneously achieving high performance and low dissipation in SIMD multiprocessor core ICs."
During the next discussion, comments were invited from specialists in systems, equipment and the like, who explained that in data centers and homes, about one-third of power is used for temperature regulation (air conditioners, refrigerators, etc), one-third is used for illumination, and the remaining third is used for everything else - computers, consumer electronics products, etc - including many products that use a lot of SoCs (system ICs).
With this in mind, it would be reasonable to be less concerned with how much power the chips consume, and to think more about how to utilize the chips to minimize excessive energy use by other equipment. In fact, at the panel discussion held in the evening as part of the same workshop, one semiconductor expert explained that even if IC dissipation were doubled, there would still be an overall improvement in power consumption if power used by air conditioners, lighting and the like could be minimized.
In response, an expert in "greening" information technology systems commented, "The power supply system was set up long before electronics came around, and hasn't changed much since. In other words, there is still plenty of room for a technological revolution, and plenty of opportunity for semiconductors."
The opportunity is not in electric power supply, but in the infrastructure, it seems. One topic raised at the panel discussion was the application of sensor networks in roadway bridges nationwide. Many bridges are said to have already exceeded their estimated 30-year life, making safety monitoring crucial, and this need could be fulfilled with sensor networks. Also, if new bridges had 100 sensors each built into them, the total number on the national level would be immense.