Mythbusters - the moon

[Rythonn]

Originally Posted by Cybsled

But no fusion reactors, thus it's not too useful ATM.

Soon!

[Valishar]

Fun Fact: Computers have already reached the limitation of the speed of light. Computer development is changing to multiple cores rather than faster processors because the actual speed of light (299,792,458 m/s or ~8.3 minutes from sun->earth) is too slow. Modern computer clock speeds are measured in Gigahertz, or billions of cycles per second. 1.0 GHz means in one clock cycle light will only travel ~33cm, 3.0 GHz would be ~11cm.

Note that those calculated distances are based upon the optimal speed of light which is moving in a straight line inside a vacuum, and neither occurs on a computer chip. At even faster clock cycles, the electrons would not be fast enough to travel from one end of the processor to the other in only one cycle, and calculations would stall.

Electrons don't move very fast in electric devices at all, in a regular light bulb the electrons move at 8.4 cm/hour, snails are faster. In a computer processor the speed is even slower since the voltage is lower. Electrical signals are transmitted at the speed of light anyway since they're EM waves traveling down the wire. I'm not to sure why they're moving to multiple cores, I think it has to do with the technical problems of miniaturization not really the speed of the clock.

[Jysin-DW]

Its more to do with the limitations of heat dissipation in the standard desktop PC. We have had 3GHz processors clocked well beyond 6GHz-8GHz range, but they have had Liquid Nitrogen(LN2) or multiple cascade phase change cooling systems.

Multiple cores are being built with the idea of parallel processing. Software needs to be written to utilize it but the idea is to take 1 large instruction and split it 50/50 between two processors. Theoretically it should take half as long to compute.

Dont know WTF youre on about speed of light and PC processors for.

*Sorry to derail*

[Malakriss]

How fast (ie: clockspeed in Hz) can a single processor in a computer be? The highest I've read is 4 GHz. What's the limit?

There is an upper and lower bound on the size of processors right now, with the speed of light limiting the upper, and the difficulties in miniaturization determining the lower. Valishar should have bolded the first two sentences in my paragraph, because he clarified how much how much slower electrons in devices and electrical signals/charges are compared to the speed of light in a vaccum.

Heat generation and dissipation is another issue associated with the higher speeds, but it's a byproduct that we are capable of dealing with rather than a limiting factor for clock speed.

I recall my Computer Architecture professor giving a lecture on the processor development and the speed of light limitation. I'll have to send him an email asking for the model # but one processor that was scrapped because the designers sped up the clock too much for the processor size, and the signals required an extra clock cycle for it to get to the other side. The problem with this is that all the instructions that usually take 1 cycle now took 2, so it completely negated all of the advantages of the higher clock speed.

6 GHz processor with extra clock for instruction times is the same or worse (delay handling) than a 3 GHz processor with normal instruction times.

[mek]

Originally Posted by Valishar

Electrons don't move very fast in electric devices at all, in a regular light bulb the electrons move at 8.4 cm/hour, snails are faster. In a computer processor the speed is even slower since the voltage is lower. Electrical signals are transmitted at the speed of light anyway since they're EM waves traveling down the wire. I'm not to sure why they're moving to multiple cores, I think it has to do with the technical problems of miniaturization not really the speed of the clock.

Newtonian physics, how quaint.

[Lefazz]

All they need to do is change the universal constant of 'c'.

Q, where are you?

[Deathwing]

Originally Posted by Malakriss

How fast (ie: clockspeed in Hz) can a single processor in a computer be? The highest I've read is 4 GHz. What's the limit?

There is an upper and lower bound on the size of processors right now, with the speed of light limiting the upper, and the difficulties in miniaturization determining the lower. Valishar should have bolded the first two sentences in my paragraph, because he clarified how much how much slower electrons in devices and electrical signals/charges are compared to the speed of light in a vaccum.

Heat generation and dissipation is another issue associated with the higher speeds, but it's a byproduct that we are capable of dealing with rather than a limiting factor for clock speed.

I recall my Computer Architecture professor giving a lecture on the processor development and the speed of light limitation. I'll have to send him an email asking for the model # but one processor that was scrapped because the designers sped up the clock too much for the processor size, and the signals required an extra clock cycle for it to get to the other side. The problem with this is that all the instructions that usually take 1 cycle now took 2, so it completely negated all of the advantages of the higher clock speed.

6 GHz processor with extra clock for instruction times is the same or worse (delay handling) than a 3 GHz processor with normal instruction times.

I thought the fastest electricity traveled through a metal was ~80% speed of light, and that was liquid mercury. Saying speed of light is the limitation, while in a pure theoretical standpoint, is true, it doesn't really consider much of the context. Besides, clockspeed is only 1 part on how "fast" a cpu can be.

Oh, as for the move to multicores, it is most definitely heat disapation. Miniaturization of transistors reaches a manufacturable limit at around 22nm currently, and I believe Intel is a couple technologies behind that(although 22nm and 32nm aren't anywhere near fully developed). What's happening now is two fold:

1. Electrons are physically passing through the transistor gates because the oxide layer that's supposed to prevent that is so thing that it can't stop them all. That creates a lot of leaky devices. There's new technologies emerging that are trying to put little strips of vacuum above the gate oxide in further attempt to stop the electron tunneling.

2. As devices become smaller and smaller they "turn off" less and less. That is, the amount of current a transistor saps while in its off state becomes larger and larger relative to its on state. That creates leaky devices too. Not sure what's being done to combat this.

Anyway, saying we can just cool the CPU's more is bad engineering philosophy, I'd rather they found out way to deal with it. The knowledge will certainly be needed going forward.

[Snugglebear]

Argument over.

[Malakriss]

Originally Posted by Snugglebear

Argument over.

Ironically, this would make more sense if the speed of light was 4.5 years from the sun->earth. Full circle! And we're spent.

[Astrocreep]

interweaved phonebook myth episode rocked. very satisfying show on both ends, even the blackpowder shark myth by the secondary team was cool.