Moore's Law musings

[BillyStubbsTears]

Years ago, Ray Kurzweil noted that Moore's Law didn't only apply to the increase in the number of transistors on a chip. He reckoned that for about 150 years, the number of calculations you can buy with $1 had doubled roughly every 18 months.

I got thinking about that today. I've just bought a 64Gb micro sd card for £17.99. From Currys so I'm sure I could have got it cheaper.

That's 28p per Gb.

Back in 1981, I bought a 16kb RAM pack for my ZX81. It cost about £50. According to Treasury figures, £1 then is worth about £3 now. So, say £150 at 2015 prices.

That's about £10,000,000 per Gb.

Something like a 30 million gold increase in the RAM per unit cost. Which is about exactly the same as a doubling every 18 months.

Think what that will mean if it carries on.

By 2040, you'd be able to buy 250Tb for £1. But the entire written output of the human race is estimated at "only" 160Tb per year.

The implications for what we will be able to do with data are just breathtaking.

[Dutch Uncle]

Theoretically yes BST, but surely these laws have limits of applicability at some stage, even it is as simple as components cannot be sub-atomic in size.

On the other hand I am sure there is a Parkinson in there somewhere - data stored expands to meet all capacity available.

As another comparison, in 1982 I had just joined a Scientific Research centre for a large international organisation. At the end of the year the travel budget, like most budgets (stupidly no flexibity between them at that time) was running out, and I was bumped off a trip to North Norway which would have involved low flying helicopter rides over the spectacular terrain. While the rest of the team were away the guy in charge of the equipment budget came to see me and said somehow there was still a lot of money left and did I have any requirements - and if we didn't spend the budget it would be reduced next year (things have changed a lot since then). I bought an Apple 3 for 28,000 Dutch Guilders - more than a Mercedes cost at that time. It didn't even have a hard drive! Actually I used it daily for at least 5 years, first came across spreadsheets  (VisiCalc) and wrote a simulation for moving traffic across North Norway terrain, so it did serve a lot of use. But 28,000 Guilders (about 3 to the pound then I think)?   

[BillyStubbsTears]

Dutch

Agreed, the exponential growth can't go on forever. A colleague of mine once noted that at the current rate of growth in academic publications, by 2140, there will be more papers published annually than there are known sub-atomic particles in the Universe.

But. The growth might well continue for a while. Kurzweil noted that the number of calcs per second per dollar has risen exponentially for 150 years through 5 or 6 paradigm shifts (manual adding machines, punched card adding machines, valve computer, transistor computers, chip computers. He reckons that, at the current level of tech, there's been a new paradigm that has emerged each time the previous one reached its practical limit. And frequently, the paradigm shift was utterly impossible to predict a decade or two before it emerged.

[tommy toes]

Can't wait for Sammy's contribution to this thread.

[Seadog Minibus]

Graphene looks to be the next step in keeping this going doesn't it?

[Mike_F]

Dutch

Agreed, the exponential growth can't go on forever. A colleague of mine once noted that at the current rate of growth in academic publications, by 2140, there will be more papers published annually than there are known sub-atomic particles in the Universe.

But. The growth might well continue for a while. Kurzweil noted that the number of calcs per second per dollar has risen exponentially for 150 years through 5 or 6 paradigm shifts (manual adding machines, punched card adding machines, valve computer, transistor computers, chip computers. He reckons that, at the current level of tech, there's been a new paradigm that has emerged each time the previous one reached its practical limit. And frequently, the paradigm shift was utterly impossible to predict a decade or two before it emerged.

Quantum computers being the next one?

[silent majority]

Graphene looks to be the next step in keeping this going doesn't it?

Possibly, still damn expensive though.

[Glyn_Wigley]

The biggest problem that restricts future developments is how you get rid of the heat that's generated.

[Mike_F]

Graphene looks to be the next step in keeping this going doesn't it?

Possibly, still damn expensive though.

Everything is damn expensive in the early days. Like the point above re early Macs. When someone comes up with a n efficient and scalable way of producing graphene they'll become very wealthy!

[Rios]

There's also been talk of going back to Germanium, which the first transistors were made from.

Diamonds are also many times faster than silicon, but I can't see that being feasible financially.

[nightporter]

Could be Photonic computing    http://www.engadget.com/2015/12/23/researchers-show-working-light-based-processor/     be a while before we see it on a desktop though.

[BobG]

If a photon based computer is on the horizon, who's gonna have the world's first photon torpedo? I want one!!!!

BobG