how will computers continue to keep up?

[oldman420]

as we aproach limits of speed in ic chips will the answer be going to ever wider busses?
will there eventualy be 128 bit and above busses to compensate?

 

[Chu]

Originally posted by: oldman420
as we aproach limits of speed in ic chips will the answer be going to ever wider busses?
will there eventualy be 128 bit and above busses to compensate?

Currently, both AMD and Intel are going the multi-core route. Also, increasing the bus speed is not always the answer, the PS2's emotion engine with its 2048-bit bus certainly doesn't have many issues with moving data around, but there is only so much prediction you can do . . .

 

[Falloutboy]

I see the futur of computers going this way. were going to see one more drop in size proubly too .06 which is proubly the max we'll be able to go to and still make any gains. multicore chips are going to be the way of the future. My guess in the next 10-15 years we'll see synthetic diamonds replace siliocon for chips, due to the fact they would be able to take extreme heat much better.

As far as busses they only can be so big until the motherboards are too expensive to make. my guess is optical busses might become a reality in the next 10 years.

 

[azel]

Four ways to speed up a delivery in anything (mail or electrons) is to:
1. increase velocity of transfer
2. decrease distance
3. send more packets at once
4. decrease required number of packets to send

Increasing speed:
You can't speed up velocity of electrons easily, but you can use faster signaling material like light. Yet building tiny integrated circuits with light is impossible at present day technology. Of course research might make it someday possible 🙂 Larger scale components are being coverted to optical signals now-a-days, though.

Decreasing Distance:
Decreasing the size of transistors and circuits is very important; half the distance can be travelled at half the time in same velocity. Decrease in size also save space. (at the cost of more interferrance and heat) A process of using water to allow higher precision of manufaturing is now getting spotlight.

Sending more at the same time:
Increasing bus is one form of this. Increasing the numbers of CPU and parallel processing has been popular for a while. The world's best supercomputers are simply smaller computers clustered to gather. Programming efficient parallel processing code is now an important topic in computer science. Intel's HyperThreading is another form of doing more at once. RAID, internet idle-time sharing projects, and dual-core CPUs all rely on this principal. Coming up are biochemical computers which use chemical reaction to calculate massive amounts of data. Quantum computing is also a form of using chemical properties of quantum physics to perform unbelievable amounts of calculations at the same time; although at less precision. Quantum computers have much spolight these days; I believe someone just created a working emulator.

Decreasing total calculation:
Compression of data is important in data transmission. New methods of compressions are always being researched. Current day computers use binary representation, but what if we represent more digits in one particle? Optical computing may be able to use different wavelengths of light to represent size of a number.

So there are lots of possible futures for computing 🙂 No worries!

 

[CTho9305]

Originally posted by: Falloutboy525
I see the futur of computers going this way. were going to see one more drop in size proubly too .06 which is proubly the max we'll be able to go to and still make any gains.

Hardly. After 65nm, comes 45nm, and that move will include changing the device gates from polysilicon to metal, which should significantly reduce leakage (and reduce resistance, so you can make longer gates or route in what used to be poly). Intel research

 

[f95toli]

Assuming someone can figure out how to use photolithography for 45 nm lines, and on a commercial scale.
Makung simple circuits using an e-beam is one thing (45 nm is easy with an e-beam), commerical productions another.

 

[CTho9305]

Originally posted by: f95toli
Assuming someone can figure out how to use photolithography for 45 nm lines, and on a commercial scale.
Makung simple circuits using an e-beam is one thing (45 nm is easy with an e-beam), commerical productions another.

I'm sure a few years ago, people said the same of 90nm and 65nm.

 

[f95toli]

True, but at least people had ideas about how it might be done back then.

As far as I know, maybe with the exception for immersion lithography, no one has been able to come up with a realistic idea for a 45 nm process (again, we are talking about commercial production, there are plenty of methods which might work for small circuits or prototypes).

In reseach labs the "limit" for photolithography is around 150 nm, anything below that you use an e-beam (not that it matters much, you need to use an e-beam to make the mask anyway). 90 and 65 nm photolithograhy is VERY specialized (how many places are there that can manage 65 nm?)