Computer idea 2

So this is an idea for a new type of computer

Have a 2d plane of devices powered and supplied data via light from a scanning laser and the fastest mems array possible. The devices return data by blinking or changing color of micro-l.e.ds

By adding multiple devices that are for diffrent tasks a scalable parallel processor could be expierimented with.

If refresh rates of mems could be timed correctly multiple imput cycles and output cycles be managed.

What are the logic gates? What does the comparison of on-off-other states?

What you are describing sounds more like an optical memory device. Like an equivalent to RAM.

The logic gates are in individual components, like small clusters that form all components. A series of small cpus and small chunks of ram (optical would be awsome) can mimic traditional hardware. I was thinking that all data could be both binary and vectors by having leds blink and change intensity of color. The only bottleneck I see is in the number of mems arrays and input streams. The refresh rate of the components will undoubtedly be faster then the mems can operate. So by offsetting the timing of the mems arrays a faster clockspeed could be achieved. One problem I see is the output speed of each component would be diffrent.......

Ok, gotcha. The calculations are done by each MEMS depending on what they are "loaded" with. So you could have a entire byte loaded with each bit on an element and then it does... whatever, evaluates even or odd, averages instead of the simple binary comparison that a conventional CPU does. The basic physical layer could be much higher "up" in the machine and closer to what computers are used for (not pushing 1s and 0s around). You could even do away with conventional programming entirely and go more to a synaptic model. Also this architecture could support pseudo-quantum computing, where wavelength represents probability.

Down side, beyond actually getting the MEMS to do your bidding, reliably, is that it will be hard to beat conventional integrated transistors for size and energy efficiency. You'd have to get at least the high UV and probably into x-rays and up to get down to the nano scale. Electrons are much better behaved.

I'd still say that a spatial array would beat planar in this system, because the components wuld have LoS on more ther components. If your LEDs are in a cartesian layout, each one wuld be ajacent to up to 8 others, or a hex array would be six. In 3 dimensions, cubits would be able to access 26 other bits, so the parallel exponent would be the cube, not the square. (In 3d, there's no regular Fuller packing that works, other than a cube.)

Another forseeable problem is that LEDs emit light, but don't sense it, which means each bit would have to consist of 2 components, and emitter, and a receptor. And finally, each is also a transducer, which converts electricity into light, then back to electricity, which causes losses, which generate heat. Enough of these bits for decent prcessing power would probably require significant cooling.

All of this assumes current technology, if you could invent components to emit, and receive signals, for instance, and optical superconductrs, it might actually be better than what we already use.

Yeah a photo cell can not be an led at the moment. I have been looking into neuroscience and the plastisity, by having the ability to pass vectors a very fast shape recognition device could be achieved.... anything that involves vectors like velocity etc.

Also with the new synapse to wire connectors could allow for humans to decode the brains comunication tricks.

LED's often are photocells and have been used as optical transmitter/recievers.

These may help;
http://www.wired.co.uk/news/archive/201 ... cient-leds
With the thermo-electric side effect, and a good superconductive substrate, it just might be possible without cryogenic cooling.

With this system you could create a optical pc that ran on ambient heat and light, or even convert solar panel waste heat into light and back into electric....

Many possibilities are open from this.

Can you cool something to superconductivity by emmiting all of its heat as light?

Another possibility is to make one of these that can withstand extreme heat to be in the heat shield of a ship so your friction could be converted into light and then radio or other emf to drive my plasma envelope idea.............

Read the artical. Normally, I would assume no, but this technology appears to do just that. I wish I had more information to figure out how it works. I suspect some feild effect in the P-N junction, but I don't know. I'm extatic, though because if we could isolate the thermo-electric effect, we could use it to recover the primary loss in all our technologies, which is heat. Also, we could line a thermonuclear reactor to directly generate electricity from it, instead of boiling water to run a turbine, at a massive net loss. That may be enough to raise Clean (AKA "Cold") fusion to positive efficiency.