The future of technology?

[GuyPerfect]

So sorry, but unless you can slow down light quite a bit, it's not incredibly feasible.

Sounds like we just need some new technology to control the flow of light at its inherent speed. It's strange that your argument takes the form of "We can't make better technology because existing technology isn't good enough for it."

[Yawg]

Kind of a zen question huh? We can't improve our technology unless we improve our technology http://www.lemmingsforums.com/Smileys/lemmings/laugh.gif" alt=":D" title="Laugh" class="smiley" />. And on a side note, you can't really "slow down" light, but you most certainly control how fast photons get from A to B, providing you have the right material between A and B. By this method, I believe our current capabilities allow minimum speeds under 100 mph. Case in point, the tech is there, it's just not widely used because it has yet to reach peak practicality and is therefore not a financially appealing alternative for all the electronic giants.

[Liebatron]

"it has yet to reach peak practicality"

Light at any speed we choose; ah...
The other thing is you'd have to slow it down all the time... Like whenever it was going anywhere it would have to be slowed down. Those would be some expensive 'wires'

Right now though each of our transistors can only store one state; 1 or 0.
A component comprised of a bit counter (without a clock component) hooked up to multiple inputs combined with another component with 4 transistors could provide 16 possible physical states with the space required to fit 4 ordinary transistors though; and it would be a pain for the software designers but you could in theory make a system X times more efficient using a non-binary method.

If you ask me that's the best way to make stuff faster. Multiple inputs, one output; it would provide for the ability to read 4 inputs at once through the reception of one output, the numerical value of which could range from 0 to 16. Of course processing would still be linear, but all data would take 1/4 as long to process due to greater compression.

[Yawg]

My money is still on quantum computers. I think thats ultimately the direction computing will take, and its just a matter of putting in the time and effort to bring our capabilities up to par.

[ccexplore]

I think at this point it might be helpful to link to a couple of Wikipedia articles as starting points for people to further explore on the topics touched by recent discussion, such as understanding the current state of research, what has been accomplished, and what challenges remain:

optical computing:  http://en.wikipedia.org/wiki/Optical_computing" class="bbc_link" target="_blank">http://en.wikipedia.org/wiki/Optical_computing
quantum computing:  http://en.wikipedia.org/wiki/Quantum_computing" class="bbc_link" target="_blank">http://en.wikipedia.org/wiki/Quantum_computing
non-binary digital computing:  http://en.wikipedia.org/wiki/Ternary_computer" class="bbc_link" target="_blank">http://en.wikipedia.org/wiki/Ternary_computer (closest thing I can find)

Right now though each of our transistors can only store one state; 1 or 0.<snip>

I have trouble understanding your proposal.  Are each of your input/output signal one electrical signal with 16 voltage levels, or just 4 ordinary binary bits?  If it's the latter it would seem that you're still dealing with a binary system?  If it's the former, how do you propose to use ordinary transistors to implement an input-output response that differentiates between 16 voltage levels instead of the typical "on/off" (N volts/0 volts)?

[Liebatron]

To do this now you'd use a system where four transistors are treated as a single output with some sort of output, but that might not actually bring an inherent gain in efficiency. Ideally you'd find a second state to change. Like find a way to control variable X that inhibits the flow of electrons the same way we control electrical on/off switches now.

EDIT: lol; apparently the link says that this method is ideally used in combination with optics. I don't like light though, so I vote we find another changeable state to work with.

[ccexplore]

I'm not really seeing non-binary digital systems having much potential to scale up.  Meaning that you'll probably have to completely re-invent and re-design your basic transistor and logic gates every time you go from N-ary to (N+1)-ary, for a diminishing gain of (N+1)/N in speed (assuming that your (N+1)-ary units do not themselves run slower due to increased design complexity).  And it seems that what you're gaining can also be gained simply by trying to exploit more parallelism out of a standard binary digital system (ie. doing more computations in parallel instead of sequentially).  While parallel computing is not easy either and has its own limits, at least it seems to be a more well-researched problem and has better potential to scale up.

This is in contrast with something like quantum computing which could actually provide exponential increase in speed for certain classes of computation processes.

=======================

It might also be worthwhile to take a step back from the Moore's Law paradigm of merely speeding everything up.  Note for example that in the human brain, signals travel from neuron to neuron at speeds thousands (millions?) of times slower than your modern PC, yet the human brain is capable of many tasks which are well beyond the capabilities of today's best computing hardware and AI software.  The gap of course is as much software as it is hardware, but it does show that speed, while important, isn't necessarily everything when it comes to computing.

[GuyPerfect]

Note for example that in the human brain, signals travel from neuron to neuron at speeds thousands (millions?) of times slower than your modern PC, yet the human brain is capable of many tasks which are well beyond the capabilities of today's best computing hardware and AI software.

Part of this also has to do with the physical size of the brain and the fact that it can multitask.

[ccexplore]

Correct.  This is part of what I alluded to when I said HW and SW.  The fact that the number of neurons in the brain are orders of magnitude higher than, say, the number of transistors in current computers, and that the software and hardware is architected totally differently from the typical von Neumann architecture of today's computers.

And it's good to remember that an airplane works quite differently from a bird's wing, so just because the brain works so differently doesn't mean that a superfast CPU with lots of memory and the appropriate algorithms, but still under the current architecture, can't accomplish the same thing.  (We're already starting to see that for example when it comes to chess playing.)  It's just meant to be an example of how the future might end up evolving in very different directions from current technological trends.

[Liebatron]

So sorry, but unless you can slow down light quite a bit, it's not incredibly feasible.

Sounds like we just need some new technology to control the flow of light at its inherent speed. It's strange that your argument takes the form of "We can't make better technology because existing technology isn't good enough for it."

Also; I should specifify; the difference here is between practical technology and pure science. We need an improvement in pure science technology's ability to slow down light before it can be turned into such a practical technological advance.