Science looks beyond silicon in search for computer speed
Dublin’s tech hub has been designated Silicon Docks but the metalloid element may be superseded by new computing materials
Silicon chip: its monopoly in the computer-chip industry may be in decline as increases in chip speed slow further
It started with Silicon Valley, then the talk was of Silicon Hills and now we have our own Silicon Docks? While the element may be as old as time itself, silicon has been a fashionable, even essential, ingredient in the manufacture of computers for only half a century.
Things change rapidly in the tech world. The rate of disruption is accelerating at an enormous pace.
So isn’t it a little short-sighted to be naming places “silicon” this or that? Granted, Silicon Valley is the Mecca for the techorati, but what happens when the material becomes obsolete? When it is superseded by some sexier, younger element? Won’t all the valleys and hills and docks distinguishing themselves as gateways to modernity all of a sudden sound, well, thoroughly un-modern?
Death of Moore’s law?
As silly as all this may sound, silicon’s monopoly on the computer-chip industry may be in decline as Moore’s law slows.
The law relates to an observation made in 1965 by Intel co-founder Gordon Moore, who predicted that over the history of computing hardware the number of transistors in an integrated circuit would double about every two years (highly purified silicon is a crucial ingredient in integrated circuits).
“This leads to a sustained increase in chip performance,” says Dr Fearghal Morgan of the NUI Galway school of engineering.
In other words, the ever-increasing number of transistors in an integrated circuit is the reason why technology is getting better and faster, leading to ever-increasing capability. “We have all witnessed this in our everyday gadgets,” he says. “But it’s slowing down.”
Currently, transistor density is still rising rapidly but the rate is decelerating. According to one report, in 1990, transistor counts doubled every 18 months. In 2015, that reportedly happens every 24 months. It’s only a matter of time before transistor density doubling slows to 36 months per generation, and eventually it will stop completely.
It’s impossible to know for sure when this will happen, but some have predicted it to stop at an effective gate length of about five nanometers by 2030 (five nanometres is about the space between 10 silicon atoms).
“One consequence of this is the increasing power requirement of modern high- density chips, and the need for significant heat dissipation,” says Morgan. “An interesting term is ‘dark silicon’, which describes a chip which cannot operate with all circuits on at the same time, since power dissipation becomes a problem. It does so much, but not everything at the same time. This is a bit like having a supercar with a radiator that cannot cool the engine at max speed.”
As Hacking the Xbox: an Introduction to Reverse Engineering author Andrew “Bunnie” Huang put it: “The implications are profound . . . Some day in the foreseeable future you will not be able to buy a better computer next year. The flash drive you purchase next will cost the same and store the same number of bits as the one you’re replacing. And you’ll have to stop looking forward to your next phone being more powerful and doing more amazing things than your last one.”
For some of us, that actually might sound quite appealing. The constant state of feeling behind the times, technologically speaking, can take its toll.
Of course, it would be naive to think an industry as powerful and profitable as silicon chip manufacturing is simply sitting back waiting for inevitable demise.
One man’s problem is another woman’s opportunity. And it is opportunities such as these that drive real innovation, particularly in a world where industry and academia grow closer every day. So the race is now on for engineers to come up with alternatives that match, or even supersede, Moore’s law’s level of improvement.
“Industry and researchers are always looking for other solutions to track Moore’s law, for example parallelising systems,” says Morgan.
“Intel’s Many Integrated Core technology uses multiple processors, like a team of processors, each performing a part of the task concurrently and communicating with the others as required. Programmers now think and code more and more in parallel, rather than depending on a single processor performing each task step by step. This style of programming can be quite complex though.”
For Bill Barth, director of high-performance computing at the Texas Advanced Computing Centre, one of the leading supercomputing centres in the US, this is mostly an economic problem for chip developers.
“Everyone is looking for ways to make computer chips cheaper, use less power and go faster,” he says. “That’s been the goal for the last 50 years of silicon research and it’s going to continue. There’s research being carried out in all sorts of areas other than silicon, such as with graphene.
“We’ve already seen lots of evolutionary steps in computer chips over the last number of years so it is just another opportunity for hardware manufacturers. But it’s also a cost. In order to keep making better chips, manufacturers will have to make strategic R&D investments.”
Who or what might save us? Well, nanotechnology may contribute something new and exciting with higher performance, lower power consumption and on smaller devices.
Other than graphene, there is research into several potential alternatives to silicon, such as the investigation into two new classes of materials - exciton condensates and spintronics - at the SouthWest Academy of Nanoelectronics in the US.
“Another advance helping track Moore’s law is 3D stacking of separate silicon dies within a single chip package,” says Morgan. “Semiconductor companies combine different types of chip technology to provide a full system on a single chip, where several separate chips would have been required in the past.”
Silicon’s future is uncertain. So maybe Ireland Inc should reconsider naming Silicon Docks something more fluid, such as Parenthesis ( ) Docks, where a new element can be inserted as the whim of modern technology dictates.