For around fifty years, the offhand statement by Intel co-founder Gordon Moore — that transistor budgets would double every year — held true.
This has been dismissed by nay-sayers more or less since its inception, whilst at the same time has held remarkably true, possibly due to self-fulfilling prophecies and our global need for more compute hardware.
Read on, however, for the real story, and how that’s all about the change.
Truth is, as I’ve stated before, that even the cheapest of budget feature phones these days has more cpu, memory and storage space available than the computer which my dad bought after much forethought and investigation around twenty years ago.
We practically give away computers that have more oomph than that long-junked PC — I know I’ve got several USB sticks for free with more than 100 times the storage density of the hard disk that graced it, and I can purchase a raspberry pi SBC for around $35 that is hundreds to thousands of times faster, depending on how you spec it — but as I said, Moore’s Law has been slowing down for the last few years, with Intel missing a few of its self-imposed tick/tock platform upgrade beats and other companies following suit. In AMD’s case, the only real competition to Intel, they’re still on 28nm fabric whilst Intel is on 14nm.
This is happening because the massive die-shrinkage which has been the norm in years past is finally coming to an end. Silicone by itself just isn’t a good enough semiconductor when the real estate scale is that low, all sorts of odd quantum effects, crosstalk, current bleed and EM field interference issues become actual issues causing otherwise rock-steady platforms to misbehave.
In short, we can’t drive silicon much faster, and we can’t make it much smaller (although 10nm, 7nm and even 5nm processes are in the roadmaps). This is the reality of Moore’s Law… for silicon.
There are, however, alternatives… such as quantum computers and memristors with their novel computing designs which threaten to completely overturn current paradigms, and of course the titular graphene, which promises to merely revolutionize computing as we currently understand it.
Graphene, in short, is dirt cheap, incredibly flexible, incredibly strong, incredibly light and amazingly suitable for using in computer chips just as soon as we’ve worked out how to make it act more like a semiconductor, and this isn’t pie-in-the-sky conjecture, this is technology which is literally just around the corner. This time next year, you’ll quite possibly be hearing of how people are printing their own CPU’s onto plastic. And if it’s not next year, it’ll be the year after that. And when anybody in their home can produce their own next-gen computers, that’s when one of the pieces powering the singularity falls into place. And even if we don’t find out how to print CPU’s onto paper/plastic, the speed and compute density improvements of graphene over silicon will propel us forwards into a new session of Moore’s Law expansion.
We’re in for some interesting times, friends.