"Software has to also start following Moore's Law," said Intel fellow Shekhar Borkar, who has responsibility for directing research into future processor technologies inside the chip firm. "Software has to double the amount of parallelism that it can support every 18 months to two years," he expanded, covering every possible base by using the most generic formulation of Moore's over-used and mis-understood 'law' possible. Further, he suggested that software devs better get on the multi-core train or risk being left behind - "For every software (company) that doesn't buy [multi-core], there is another that will look at it as an opportunity," he threatened.
Intel is renowned for using Gordon Moore's processor prediction in the most
useless diverse ways. Hacks at Intel's Developer Forum shindig are known to keep a count on the amount of times
Moore's Law is dragged out to justify some completely random trend. This has the benefit of swinging both ways - the
appearance of Moore's Law everywhere gives a headline suggesting Intel is spouting lots of the usual rubbish and making
up appalling post-justifications for trends it's trying to predict; a lack of Moore's Law references can lend itself to
a headline of 'Moore's Law dead', thus attracting ire from the chip giant.
More parellism is needed because both Intel and DAAMIT are focusing their hopes of future success on creating more cores, not upping processor clock speeds, in a bid to do more simultaneously. Cnet hits the nail on the head by suggesting that the pair "are delivering processors that have multiple brains, or cores, rather than single brains that run ever faster." God help us. ยต