The clock has been ticking on gigahertz for years. One the face of it, time is up and gigahertz now rank right down there with off-die caches at the bottom of the list of priorities for new chip designs at Intel and elsewhere. But speed is still important under the covers. At Intel's Developer Forum in San Francisco last week it took the company all of five minutes to invoke Moore's Law, and within a decade, we heard, the steady march to Moore's tune will bring teraflop performance right onto our desktops.
While the hard count of Gigahertz is out, multi-core is in; there is no architectural limit, we heard, on building chips with up to 4,000 cores. Microsoft may have a problem running Windows on more than 64 cores, but the magic of virtualisation offers a way around even the limitations of Redmond.
That's par for the course, but at this IDF there was also a lot more talk of what technology can do for parts of the world not normally concerned with the fastest processor or the latest storage technology.
Professor Eric Brewer, who runs Intel's lab at Berkeley, is likely to be as interested in the economics of goat farming as chip yields. The fact that Intel is struggling like many other tech companies to figure out what technology is appropriate for the developing world was highlighted when he pointed out that car batteries held up by others at Intel as an ideal power source for PCs in the developing world, are in fact nothing of the sort. After all, they are full of nasty stuff you wouldn't want dumped in your own back yard.
Of course, we should not fool ourselves that we are seeing a new, altruistic side of Intel. This focus on emerging markets and other new, human areas such as health is driven by a search for new and untapped budgets. Healthcare is famous for the amount of money spent on tech, and the near-infinite more it could spend trying to get the first lot to work. And, as Brewer pointed out, the myth that people in developing nations have no money to spend on stuff (particularly appropriate technology) is largely just that — a myth.
It is, however, heartening to see people within companies such as Intel realising that they still have little idea what that appropriate technology is. The gigahertz that matter are probably in the radio link, not the chip; it probably is smaller, dust- and humidity-proof, and probably doesn't require much power to run. But really, only the people who would use it can say. Intel is trying to listen.
Particularly in the developing world, the right technology is essential. A million PCs running off a million car batteries may indeed be wholly inappropriate, but in the absence of a proper, appropriate technology, the next best — or cheapest — thing will be used, and the longer it takes for the combined tech companies of the world to come up with the right technology, the more damage will be done to environments that can least sustain that damage. For the people who live there, the clock most certainly is ticking.