If you paid any attention to the news this week, you might have heard a little something about Moore’s Law finally breathing its last, exasperated breath. Of course, Moore’s Law has been declared “dead” several times over now, only to be resurrected by a new type of silicon, a refreshed diode manufacturing process, or the great white hope of quantum computing.
So what makes this time different?
First coined back in the earliest days of computing, Moore’s Law suggests that the amount of available computing power on any given chip doubles once every 12 months. This law has remained as a constant until recent years, as manufacturers like Intel and AMD have struggled against the materials used to print processors (silicon), and the nature of physics itself.
The issue that chip makers are facing lies in the world of quantum mechanics. For most of modern computing history, Moore’s Law was a constant, a reliable way that both manufacturers and consumers could chart how powerful they could expect the next line of upcoming CPUs would perform, based on the technology of their predecessors.
The less space between each transistor, the more of them you can fit on a single chip, which increases the amount of available processing power. Each generation of processor is graded on its manufacturing process, measured in nanometers. For example, the 5th-generation of Intel Broadwell processors feature logic gates that are rated at “22nm”, which designates the amount of space available between each transistor on the CPU’s diode.
The newer, 6th-generation Skylake generation of processors uses the 14nm manufacturing process, with 10nm set to supersede that around 2018. This timeline represents the slowing of Moore’s Law, to a point where it’s no longer consistent with the guidelines that were originally set for it. In some respects, this could be called the “death” of Moore’s Law.
Quantum Computing to the Rescue
Without getting too technical, quantum tunneling uses tunneling transistors that can harness the interference of electrons to provide consistent signals at small sizes, while spintronics uses the position of an electron on an atom to capture a magnetic moment.
It could be quite some time until either of these technologies are ready for full-scale commercial production however, which means until then, we might see processors take a different turn for low-power consumption over high-horsepower.
For now, companies like Intel have said that instead of prioritizing the need for raw power or clockspeed, processors will need to start actually rolling back how much power they use in favor of increased efficiency.
This is a shift in processing technology which has already been happening for a number of years now thanks to smartphones, but now the pressure to include devices like those under the umbrella of the Internet of Things in that same category is changing the way we think about CPUs as a whole.
It’s predicted that as we begin implementing more technologies that utilize quantum mechanics, mainstream processors will have to slow down for awhile before they can catch back up, as the industry grows through the transition phase between the two generations of CPU-printing technology.
Of course, there will always still be a demand for processors that can run games and applications on desktop PCs as fast as possible. But that market is shrinking, and low-power, ultra-efficient processing will still be the favored choice as more mobile and IoT devices start to dominate the market as a whole.