We have transistors, and we can build different types of transistors.

So, everyone thinks about transistors today.

So people were building computers long before transistors, and people were thinking about computer architecture long before transistors.

Well, lets say all of sudden, you get a big bump in transistors, you get twice as many transistors.

Let's think of these transistors as little switches.

Now, this is starting to happen in transistors.

And now of days we, of course, have transistors.

Whole different paradigm transistors came out of the woodwork.

Right now we manufacture about 10 billion transistors every second.

So the most likely, the most common component put on here is transistors. And you can make all sorts of stuff out of transistors.

We went to the fourth paradigm, transistors, and finally integrated circuits.

A lot of times what happens if you look at this graph is what's happening is you're getting more transistors, but those transistors are not necessarily exponentially faster.

Well, before we had transistors we had little glass tubes that actually looked

Now, in a few years time, by 2015, we will shrink transistors so much.

And these transistors are behaving essentially just like ion channels behave in the brain.

So you can have switches that turn, change switches, similar sorts of ideas to transistors.

He said he can get twice as many transistors for a certain amount of dollars.

Most of you probably have 100 million or so transistors in your pocket right now.

A quadrillion transistors is almost the same as the number of neurons in your brain.

Transistors are getting smaller to allow this to happen, and technology has really benefitted from that.

Or you can think of it as just that the transistors are just getting cheaper and cheaper.

And what is really important here to note is even if you have better and better transistors.

The way that silicon behaves, the fact that you can build transistors, is a purely quantum phenomenon.

Likewise technology constraints push up, so as we go to smaller and smaller transistors, we'll say the smaller and smaller transistors let's say go faster, but the wires go slower, well that's going to influence your computer architecture.

And it's just an observation about how the manufacturing of chips with transistors on them tends to go.

You can make both p and n type semiconductors, which means you can make transistors out of them.

170 quadrillion transistors, 55 trillion links, emails running at two megahertz itself, 31 kilohertz text messaging, 246 exabyte storage.

The guy who founded Intel said that every two years, the number of transistors on a computer chip doubles.

Now, what's happening is that as transistors are getting smaller and smaller and smaller, they no longer behave like this.

And people have also sometimes taken this to mean that you get twice as many transistors on a chip every year.

And more than that, it's a confection of different ideas, the idea of plastic, the idea of a laser, the idea of transistors.

And we take it for granted now, that each of these machines has billions of transistors, doing billions of cycles per second without failing.

Moore's Law was just the last part of that, where we were shrinking transistors on an integrated circuit, but we had electro mechanical calculators, relay based computers that cracked the German Enigma Code, vacuum tubes in the 1950s predicted the election of Eisenhower, discreet transistors used in the first space flights and then Moore's Law.

But by the teen years, the features of transistors will be a few atoms in width, and we won't be able to shrink them any more.

So now the neurons are represented by little nodes or circuits on the chip, and the connections among the neurons are represented, actually modeled by transistors.

And what the law says is that every 18 to 24 months, approximately, the number of transistors which can be fit onto one of those chips doubles.

And a current integrated circuit might have in each one of these chips something like a billion transistors, all of which have to work perfectly every time.

So, what computer architects have to do is we need to figure out how to take buckets and buckets of more transistors and turn them into more performance.

And by Moore's law, the manufacturing of that chip... because there's more transistors that fit in there, for pretty much the same price they can offer more and more capacity.

They're faster, so you've got exponential growth in the speed of transistors, so the cost of a cycle of one transistor has been coming down with a halving rate of 1.1 years.

And, as you know, if you can turn a given algorithm into a fixed function and just, hard code into transistors, you get much better performance like texturing.It's one of those things where.

What that means is that you can imagine a machine that's consistent with all the laws of physics in which every one of those 10 million, 100 million transistors in your pocket was simultaneously on and off.

Now if you could build one of these machines that took advantage of the superposition that let all the transistors in the circuit be in multiple states at the same time it becomes a very easy problem.

So, you could sorta take that two ways. It might mean that for each year when I make a chip after it doubles, well, now I can make that same chip but I get twice as many transistors.

If you've heard that term before, what it is, is we're trying, Moore's, Gordon Moore said that every eighteen months to two years, you're going to get twice as many transistors which, you can have for the same amount of dollars.