Intel’s co-founder Gordon E. Moore predicted back in 1965 that processing speed would double every 18 months--Moore’s Law. Now however, Intel’s Executive Vice President William Holt is saying that Moore’s Law is coming to an end. You may think that this is expected, that as technology is getting faster and faster we were bound to break through this half-century old law. Than you may be surprised to hear that the “end” is not because of greater increases, but actually speed decrease.
At the International Solid State Circuits Conference in San Francisco February 1st, while discussing new technologies, Intel’s Holt said that we are going to be seeing some major changes. Intel has always showcased new chip generation’s energy efficiency. But the amount of silicon transistors they can pack onto a chip--what Moore based his would-be law on--is reaching the industry's limit. Intel’s focus on furthering power efficiency, coupled with the diminishing 50 year old law, are the two players at hand as Intel shapes its future. Holt stated, “The best pure technology improvements we can make will bring improvements in power consumption but will reduce speed.”
While some will see this as a step forward for “pure technology improvements”, I for one am glad to have purchased a Skylake i5 processor in time for a computer I just built and didn't wait for their next processor family. Would you hold out for the iPhone 7 if you knew its performance would be in 5s class just to save some power? This is the kind of change Intel is proposing. Holt says we can expect not only a end to Moore’s Law, but an actual rollback in performance.
Energy consumption may take a backseat to most over a processor's ability to perform, but where energy savings becomes more pressing is in processing at its largest scale. “The carbon footprint of data centers...is growing at an alarming rate. And the chips needed to connect many more household, commercial, and industrial objects...will need to draw as little power as possible to be viable,” Holt stated at the Conference.
Big processing and data centers will probably feel this transition the most, and the rollback in performance will linger for the next few years according to Holt. Most people will probably not notice a difference, but at its largest scale I’d think, big companies certainly will suffer until whichever technology Intel chooses to pursue matches what we already have in terms of performance. Hopefully companies that are dependant on huge amounts of processing have just recently gone through a refresh cycle, or will soon before Intel rolls out their underperforming chips, or that they project to save enough money on energy that they can just purchase more processors. But all of this remains to be seen, price, performance, and energy.
Intel’s decision on what technology to invest in for the future carries a lot of weight and risk. They are considering Quantum tunnelling, or spintronics, computing technologies that respectively involve particles and electrons, and will have to re-tool all of their manufacturing plants accordingly. “The new technology will be fundamentally different.” Holt said. Their decision will ultimately decide where the future and next phase of computing will lie. It will be interesting to see how much more energy efficient these chips will be, and how long it will take for the new technologies to really take off. This is a necessary transition but one that will first have to take a step backward. The first cars were slower than the horse and buggy for years too, but still, I think I’d wait for the Model-T.