How many transistors will it take to simulate one transistor?

[Braznor]

A single real transistor may contain a few hundred to few thousand atoms or millions. So how much computational power will it require to calculate the quantum simulation of one real transistor?

 

[DownTheSky]

Yes but how many transistors will it require to simulate the transistors needed to simulate one transistor?

 

[Printer Bandit]

Let's check with the aliens who gave us the technology.

 

[lamedude]

I think Pong is playable on DICE. So 10 million?

 

[Thebobo]

A single real transistor may contain a few hundred to few thousand atoms or millions. So how much computational power will it require to calculate the quantum simulation of one real transistor?

Zero

 

[BUTCH1]

Question is worthless, a transistor is an electronic "switch" that is either in one of 2 states, on or off, a $39 celeron CPU will have over 400,000 of these, why would you want to "simulate" one..

 

[John Connor]

How many transistors are in the OP's head?

 

[Cogman]

Question is worthless, a transistor is an electronic "switch" that is either in one of 2 states, on or off, a $39 celeron CPU will have over 400,000 of these, why would you want to "simulate" one..

Nope. Transistors are analog devices (infinite number of states).

For example, Amplifiers use some of the properties of transistors to take a source input voltage and amplify it by some factor (or attenuate depending on what you are shooting for).

I would suggest reading more about how they work here.

http://en.wikipedia.org/wiki/Bipolar_junction_transistor
http://en.wikipedia.org/wiki/MOSFET

They can be quite complex depending on the model used to represent them.

 

[Thebobo]

Nope. Transistors are analog devices (infinite number of states).

For example, Amplifiers use some of the properties of transistors to take a source input voltage and amplify it by some factor (or attenuate depending on what you are shooting for).

I would suggest reading more about how they work here.

http://en.wikipedia.org/wiki/Bipolar_junction_transistor
http://en.wikipedia.org/wiki/MOSFET

They can be quite complex depending on the model used to represent them.

In actual use transistors can be either switches or amplifiers.
http://en.wikipedia.org/wiki/Transistor

 

[Cogman]

In actual use transistors can be either switches or amplifiers.
http://en.wikipedia.org/wiki/Transistor

Those two cases cover a lot of ground. My point was that you can't simply represent a transistor as a simple switch. If you want to simulate a transistor, it takes a fair bit more than representing it as on or off.

 

[Markbnj]

Those two cases cover a lot of ground. My point was that you can't simply represent a transistor as a simple switch. If you want to simulate a transistor, it takes a fair bit more than representing it as on or off.

But the question is whether it rises to a level of complexity, in terms of numbers of variables and calculations needed to update their state, that would require a notable amount of computing power? My knee-jerk answer was also "No," but maybe I need to learn more about transistors.

 

[utahraptor]

Not necessarily as many as you think. The simulated reality we exist in could be running at a speed much much slower than that reality and so on down the line.

 

[SunnyD]

 

[Cogman]

But the question is whether it rises to a level of complexity, in terms of numbers of variables and calculations needed to update their state, that would require a notable amount of computing power? My knee-jerk answer was also "No," but maybe I need to learn more about transistors.

Not a notable amount of computing power, no. In fact, we have some pretty good computer simulations for this sort of stuff already. The amount of computational power needed depends on how accurate you want the representation to be. There are simple modeling equations that cover most problems, but once you start talking about things like the physical layout of the transistors and their proximity to each other, then things get really dicey pretty quickly.

That currently isn't viably simulatable. The likes of Intel and AMD instead use rules of thumb guidelines for transistor placement rather than relying on fully proven out simulations to figure out just how close you can smoosh those things together.

Of course, that is mostly for the complex logic circuits (which are taking up a very small portion of diespace) I've got little doubt that caches have received a much larger portion of computational power to optimize size on the die. Especially since cache structure is pretty simple all in all.

But I digress. The answer is, you could accurately simulate a transistor with any modern computer (and probably most graphic calculators). The complexity increases exponentially (maybe even factorially) when you start to introduce new elements.

How many transistors are needed depends. Probably enough to get a processor put together that can start doing mathematics.

Or you could take the dumb route and say "one", since transistors tend to behave like transistors oddly enough.

tl;dr Complexity of simulation increases as the simulation become more accurate.

 

[Markbnj]

There are simple modeling equations that cover most problems, but once you start talking about things like the physical layout of the transistors and their proximity to each other, then things get really dicey pretty quickly.

Yes, but that is all the difference in the world . It's the difference between modeling a particle, and a particle system.