Various questions...

[zidane1strife]

Computing:

Whatever turned out of this intriguing cooling solution(air cooling as fast as water):
nanolightning cooling

AMD:

I've heard that the dual channel feature requires same brand same model ram cards/thingies. Which makes sense. But I've also heard they've to also be from the same factory lot/batch, or you may've problems. Is that for real?

Quantum:

When Electrons move to higher/lower energy states it is said they're receiving/emitting a photon. I've heard that this is always the case, is this so?

What's the deal with the "anything can happen but with a practically impossible probability" according to quantum mechanics saying, many physicists use in the media? Is it for real? That's can a car suddenly turn solid gold? can an ice cream cone suddenly disappear from your hand?

Wifi:

I know the army's thinking about using smart dust, millions/billions of wireless sensors on the field, on top of millions of probably wifi equipped soldiers. Now say an allied nation goes and deploys that too(say the military doesn't want to share). Now say a few allied enemy nations do the same too.... not to mention the avg. population wifi use, say ubiquitous computing(with even the bubblegum wrappers having built in chips.), assuming it's urban... What happens?

How many wireless devices/sensors can you have in a small room/space/battlefield without problem? with low energy consumption(remember wireless tiny sensors can't consume much if they're gonna be operating for any significant amount of time)?Millions? Billions? Trillions?


Could one for example use high bandwidth wireless on a future super computer? (say build a billion/trillion cheap low-power SoC chips with built in wireless). I mean If you could use wireless, while perf might suffer, it would be made easier to set it all up, no? You'd just keep buying more and put them right next to the others(assuming apt cooling), if the s/w had the ability to bypass malfunctioning/non-repairable units, it'd be quite easy to assemble and maintain, I'd pressume(just keep adding more.) or at least manageable.

What about antenna range/size/power scaling, I mean say for ubiquitous computer, could a bubblegum wrapper transmit a decent wifi signal, would it have to go optical(laser based)?

 

[zidane1strife]

You can answer some/one of the questions if ye like. replies need not answer all questions.

The last one is the one I'm interested in the most, I'd like to hear an answer to that one.(yes I've already googled several books worth of data, but I could use a simple yes or no and save my time.). I know radio antennas can receive even km~ long wavelength signals, while being but an infinitessimal fraction of such. Can smaller (cm-mm) wavelengths be dealt with antennas a fraction of such wavelengths say micrometer scale or nanometer scale ones? I've heard of some but I'm not sure. what's the limit on smart dust/mote size antennas for allowing wireless communication between the various units?

 

[arjan de lumens]

Computing:

Whatever turned out of this intriguing cooling solution(air cooling as fast as water):
nanolightning cooling

Having postive ions flying all over the place sounds like it could result in charge accumulation in places where you really don't want it - it doesn't take a whole lot of charge to damage a transistor through a CPU pin. Also, with a distance like 10 microns being mentioned, I would be a bit worried about having the device gummed up with dust.

Quantum:

When Electrons move to higher/lower energy states it is said they're receiving/emitting a photon. I've heard that this is always the case, is this so?

Yes. You can also produce/absorb photons by moving around energy states of other charged particles, such as protons.

What's the deal with the "anything can happen but with a practically impossible probability" according to quantum mechanics saying, many physicists use in the media? Is it for real? That's can a car suddenly turn solid gold? can an ice cream cone suddenly disappear from your hand?

The basic idea is that for every particle, rather than an exact position/velocity at all times, you only have a "wave-function", that is, a mathematical function that describes the probability of the particle of being at a given point at a given time. If you want to dig deeper, the relevant concepts would be Heisenberg's Uncertainty and the Schrödinger Equation. The funny part comes when you realize that that the wave-function for a given particle takes on an extremely small but nonzero value for nearly every possible position in the universe, and that each of the ~10^80 particles in the known Universe basically has its very own wavefunction. As it follows, you can take the configuration of protons, neutrons, electrons etc that make up the car in your example, describe another configuration of those particles that would make up a solid-gold car, and show that there is a small but nonzero probability that the particles will rearrange themselves into that configuration. AFAIK, the probability that it will happen to a specific car in your lifetime is however on the order of roughly 1 in 10^(10^30), so don't hold your breath. The phenomena associated with "anything can happen" more often manifest themselves at smaller scales, with the classical example being radioactive decay. The whole thing sounds freaky as hell, but the underlying formulas are fairly precise and have so far been in agreement with all known experiments (which are probably numbering in the thousands by now).

 

[Frank]

Could one for example use high bandwidth wireless on a future super computer? (say build a billion/trillion cheap low-power SoC chips with built in wireless). I mean If you could use wireless, while perf might suffer, it would be made easier to set it all up, no? You'd just keep buying more and put them right next to the others(assuming apt cooling), if the s/w had the ability to bypass malfunctioning/non-repairable units, it'd be quite easy to assemble and maintain, I'd pressume(just keep adding more.) or at least manageable.

What about antenna range/size/power scaling, I mean say for ubiquitous computer, could a bubblegum wrapper transmit a decent wifi signal, would it have to go optical(laser based)?

Step by step:

Many stores put paper strips that contain an aluminium foil antenna for preventing theft into their products. This works by having a transmiter that sends pulses of just the right frequency all around, and listening if they get a return signal. Because when that antenna picks up the signal, it is transformed into a tiny electrical charge, that transmits a pulse itself when the energy is released.

So, you can power stuff by transmitting a radio signal (generally in the GHz range) onto it. That's also how your microwave works. The energy is absorbed, and released again in a different form. Heat for food, a radio signal for those anti-theft devices, and you can also use it to power a chip for a short while.

Which is how radio tags work. Put two of those aluminium foil antennas on top of each other with a very thin isolator in between. That way, the foil doesn't only function as an antenna, but as a capacitor as well. And when you attach a chip, the energy captured is "slowly" released through that chip, instead of al at one go as a radio signal, thereby powering it for a moment.

Of course, we're talking very tiny amounts of electricity here, but as anyone who owned a digital clock for many years and wondered how many years longer the battery will last, small chips can function on very minuscule amounts of electricity.

But transmitting a signal takes many orders more energy than computing something. So, if you want the chip to send a return message, you have to compute that message, and then pulse the remaining amount of energy in a binary stream.

To do that, you just short the foil capacitor for very short amounts of time. That's basically how the very first radio transmissions worked as well. Until it's depleted. And this is how RF ID tags work, like they want to use for identity papers, money and other stuff you want to be able to identify as simple as possible. Some companies use them to track their inventory, with mixed results.

To build a useable computing system out of that, there are basically two related problems: you can only have one of those chips send at any one time (serial), because otherwise the signal is lost in the noise of all those chips transmitting at the same time, and if you use many of them, the last one needst to be able to store it's charge (power) until it get's it's turn.

The last problem is easy to solve, by just using a transmitter that sends a continuous microwave signal of low power over the whole area. If you're using batteries and want to cover a substatial area, you might prefer to use pulses. But that first problem makes it a bad fit for creating large amounts of parallel computing power, as you have to read them serially. Although they can do substantial computations, if you pulse them fast enough.

All in all, that makes it a great system for serially checking many remote sensors a short distance away, and it even allows the transmission of video, as long as you keep a directional microwave transmitter pointed at it. And the batteries won't run out (a major problem nowadays with field equipment), as they don't have any.



If you're only interested in the parallel use of all those things, you have to limit it to stuff where all those chips run independent and only talk to their direct neighbours (extremely low power), to prevent the noise. A wall-sized television might fit the bill, but that will take some time (probably decades). Theoretically it's not all that hard, but the whole technology is still in it's infancy.

 

[zidane1strife]

All in all, that makes it a great system for serially checking many remote sensors a short distance away, and it even allows the transmission of video, as long as you keep a directional microwave transmitter pointed at it. And the batteries won't run out (a major problem nowadays with field equipment), as they don't have any.



If you're only interested in the parallel use of all those things, you have to limit it to stuff where all those chips run independent and only talk to their direct neighbours (extremely low power), to prevent the noise. A wall-sized television might fit the bill, but that will take some time (probably decades). Theoretically it's not all that hard, but the whole technology is still in it's infancy.

I was thinking more of nanotechnology, particularly drexler's old pre-nanofab-model molecular assemblers and how they might communicate. I've serious doubt about that type of nanotech being versatile enough to work outside a specialized environment(especially when it comes to things like communication, energetics, fuel-acquisition and waste disposal issues with regards to surface area to volume ratio, as most of it's supposed to be relatively inert diamondoid.).

 

[Blitzkrieg]

With dual channel is it for a skt 939 or an older athlonXP with nforce2?

Never had a problem mixing and matching on an nforce2 tho heard others have. I dont know much about how picky the athlon64 memory controller is. I know it is more "flexible" in the latest E revision but just how much I am not sure.

 

[Frank]

I was thinking more of nanotechnology, particularly drexler's old pre-nanofab-model molecular assemblers and how they might communicate. I've serious doubt about that type of nanotech being versatile enough to work outside a specialized environment(especially when it comes to things like communication, energetics, fuel-acquisition and waste disposal issues with regards to surface area to volume ratio, as most of it's supposed to be relatively inert diamondoid.).

Well, I don't think there is anyone who has even a remotely practical idea about how those might be able to function at all, let alone how you could build them. If you want things like that, you should use biological constructs. Build the genes of the construct you want, inject it in the right host cell and see what happens.

 

[zidane1strife]

Well, I don't think there is anyone who has even a remotely practical idea about how those might be able to function at all, let alone how you could build them. If you want things like that, you should use biological constructs. Build the genes of the construct you want, inject it in the right host cell and see what happens.

Not enough versatility, but I believe my knowledge has opened my eyes, I know how to get a versatile enough design. Question is like any other big project(going to the moon, mega-engineering projects), can I figure out a way to do this with my own funding or will I have to search for external funding, which seems ever more likely(darpa, industry, etc.). Right now I' evaluating possible paths, and seeking the path of least resistance.

With dual channel is it for a skt 939 or an older athlonXP with nforce2?

Never had a problem mixing and matching on an nforce2 tho heard others have. I dont know much about how picky the athlon64 memory controller is. I know it is more "flexible" in the latest E revision but just how much I am not sure.

nforce 4, athlon 3200 90nm non-64, iirc.
corsair 1GB ddr 400(I think)

 

[StefanS]

Yes. You can also produce/absorb photons by moving around energy states of other charged particles, such as protons.

No that's not always case (unless we're talking QFT here). If an Augereffect takes place in an atom, instead of emitting an X-Ray cascade, the energy is transfered to the so-called Auger electron which is then emitted from the atom.
In fact, for low Z atoms this is more common than the respective X-Ray photon emission.