E&T found this year's International Solid State Circuits Conference overshadowed by the wider financial gloom.
The International Solid State Circuits Conference (ISSCC) has two main levels. The first reflects its technical foundations - it shows you where the semiconductor industry plans to go in the next five years. The second is strictly business - big players show off their latest benchmark chips, typically top-of-the-range microprocessors, memories and silicon for mobile consumer devices that will appear over the next 12 months.
That is how it normally works. This year's ISSCC may, as ever, have taken place in February at San Francisco's huge Marriott hotel, but the global economic crisis hung over the event. There was no need for overflow rooms, and even the keynotes were more sparsely attended than usual. The flavour of what companies said had unmistakably changed.
A long-standing joke is that ISSCC's opening day is marked by a front-page Wall Street Journal article on Intel's latest MPU. Another record-breaker - the company's 2.3 billion-transistor 45nm Xeon server chip - was on show, as were details of the incoming Nehalem architecture on which Intel's next immediate generations of chip will be based. But the company did not seem as interested as usual in pushing these types of innovation.
The tone was set before ISSCC at Intel's pre-briefing, featuring fellows Mark Bohr and Krishnamurthy Soumyanath. Sure, the microprocessors were important, but Bohr (who also gave an excellent keynote at the conference) was more focused on the big-picture issue of maintaining the exponential performance curve set by Moore's Law.
Soumyanath, for his part, seemed more excited by Intel's innovations in power management, graphics and analogue-to-digital converters (ADCs).
In that last area, the company has ported the multicore concept from processors to ADCs to produce devices that can handle a massive throughput of data. In short, there is some very cool stuff in there, and it has an important economic subtext. At a time when the downturn is forcing stock market investors to reconsider what 'blue chip' means, Intel must demonstrate that it is not 'just' a micropro-cessor company with products that could be commoditised (a charge it has faced even in good times) but a technology powerhouse developing the products that will trade the global economy out of the doldrums.
The medical strand of ISSCC continues to produce fascinating technology. In the memory world, SanDisk and Toshiba have hit a new density of four-bits-per-cell, promising 64GB flash cards before the end of the year.
NEC, meanwhile, showed off a flexible and closely stacked combination of memory and logic that delivers system-on-chip performance (i.e. integrated into a single piece of silicon) at system-in-package prices (i.e. stacked die with ultra fast interconnects and flexibility). Indeed, NEC's innovation in terms of 3D semiconductor packaging, architecture and interconnect was emblematic of how economics dominated ISSCC.
To be blunt, many companies did not really show up. It was noticeable that some global players only allowed staff presenting papers or those based in Silicon Valley to attend the event (and one Top 10 chip company was said to have no one at all at the conference).
Then, among the powerful research institutes, it was noticeable how some 'themes' did not advance at ISSCC although they were expected to. Officially, there was nothing to say this February; unofficially, corporate sponsors for the research were said to have slashed or pulled funding and work had slowed or stopped correspondingly.
As a result, the technologies that dominated ISSCC were those that promised to maintain or improve performance, but with a still-closer emphasis on cost. At the same time, one could also see chips that were designed at extending or maintaining markets that could be seen as being at risk.
Taiwan's MediaTek presented an integrated backend SoC for BluRay disc players, bringing together content management and decryption, video decoding, HDMI and other functions. The objective is to reduce the cost of these high-definition DVD players that have still to capture the mass market or dip below the $150 barrier.
Semiconductors have always been about cheaper, faster, smarter - and historically, it is a challenge that they have met. Nevertheless, no one could have left this year's ISSCC lacking the impression that these challenges have intensified. As to how prepared the silicon industry is step up its game, there was certainly evidence that it can - but also a suggestion that not everyone can. That, perhaps, was the conference's worrying twist.