Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 7 years ago.
Improve this questionPrototype working ternary device
Quantum computing with qbits and so on is one thing, but what exactly do we stand to gain from a CPU which works on a ternary basis, where e开发者_高级运维ach 'bit' is ether yes/no/maybe (or true/false/filenotfound)?
Is it simply an academic exercise or could it change processor design?
What practical use does increasing the number of bits have in general in computing? You get a larger address space and faster operations (like adding or multiplying, the bigger the word size, the faster the code, assuming you use the entire word size, because you only pay the register transfer cost once instead of twice or 3 times).
Increasing the "size" of a bit provides a similar bonus, you can reduce the word size and still maintain the same word range. In addition your computations (might) be cheaper since you apply your algorithm for fewer "digits" (depending on how expensive multiplication is in base 3).
It provides better density (for parity in manufacturing).
That they provide three states per 'bit' does not make them 'yes no maybe' any more than it means "true false file-not-found" - its an application level thing to decide how to interpret and label those three states, but they are not 'fuzzy' approximate states, they are absolute and exclusive.
Ternary components would actually be compatible with binary CPUs - the key distinction is if they are digital or analogue, not if they are binary, ternary or other based. Its a simple hardware problem to convert one base to another and provide interfaces in arbitrary bases - it won't require a new CPU architecture to have some memory that happened to be ternary, for example.
Actually, not 0|1|2. It would be -1|0|1. I think the possibilities brought forth in introducing a ternary system could very well change computing as we know it. Most computers operate in binary...i.e. Yes/No. Human minds operate in ternary... Yes/No/Maybe...Could this be the key to true artificial intelligence??? I think computers could behave more like man if they were given the option to doubt :)
Boolean Logic is intrinsically binary and the basis for AND/OR gates, but Dunno+Boolean (Doolean) Logic can have some utility - in Turing Machine terms, a machine that stops and says Accept or Reject is Boolean, but in general a machine can keep running and be in an interdeterminate or don't know state (loop is used in theory of computation to indicate this, but don't know and don't care states are used in optimizing circuits - finding a minimal circuit that handles what you are interested and gives either True or False for the don't care states. Circuits can also be in indeterminate states, and determinacy/indeterminacy propagates so it is easy to generalize Boolean truth tables to Doolean truth tables - much like the way NaN propagates in Floating Point arithmetic.
In terms of arithmetic operations, all numeric operations can be carried out in any base, although some have very simple forms in binary (and there are also problems that suit other bases, including that of the natural logarithm). Incidentally there is no need for the three states to be 0, 1, 2 or -1, 0, +1 or even have equal differences (could even use imaginary or transcendentals values, like i and pi or e).
The primary disadvantage is all that we have invested in binary logic machines.
The primary advantage is efficiency in that one wire or one capacitor can as easily distinguish +1, 0, -1 as just 0,1 - negative voltages are just as real as positive voltages. Then yes, there are quantum possibilities (qutrits), optical possibities (e.g. polarization), etc. But focussing on the dynamic memory and the advantages of balanced circuits, circuit density per bit would be increased over 50% (2 trits has 9 states, 3 bits only 8 states), and power consumption per bit would likely be halved, and computation time would in principle reduce by a third per bit per cycle.
i'll give a good answer to your question. First of all, in order to knock transistor's energy into a ternary transistor, you first need a power supply capable of 3 types of current. Commonly you have only 2. Backwards and fowards. Negative and positive, or 0 and 1. With 3 you simply add another direction to the current. Adding an aluminum pole to an already existing zinc (negative) and copper (positive) battery can achieve this goal. Aluminum is used because it is the opposite of brass. Brass is the mixture of zinc and copper melted together.
Ok, before I can expound upon that I want to make it easier for you to understand the fundamental nature of ternary. As an example, remember that in programming a 3d polygon, (THREE dimensional, remember that) you can't do it without having at least 3 points. But you could code something that uses only 2, but then quickly flashes to another 2 points with one of the points being where the previous rendering showed one. An imitative, fake polygon it would be, but a true polygon obviously as you know needs 3 points. ANOTHER EXAMPLE:Take the RGB color scale for instance. Same thing. You can't make whatever color you want without 3 base colors. Many things operate in threes and therefore you can almost see a certain necessity for a ternary computer. So to prove my earlier conjecture true, look at purple. It is a mixture of red and blue. So what is purple's opposite? GREEN! This is because when you are working with a properly separated color scale, all you do to find the 3rd color you need is to add the 2 you already have and then invert your result. In this case, it is green.
Now going back to the point - with a ternary computer your power supply needs that 3rd current to properly change from your initial transistor to another in a 3-option circuit/intersection. So basically you have a left, right and up (for instance) option at any given transistor. It has 3 paths it can take. You can also add more with this technology but anymore than 3 paths to each option is redundant. So in order to move it to a certain transistor you have to manipulate the electrical current slightly. In binary this is done by tricking the transistor that is charged to have both 0 and 1 for just one moment, while the binary processor or binary logic chip in question is designed to make this easier by having the correct pathway/node-switching structure. So what happens is, instead of going straight ahead which would be the most logical step when turning on the next transistor, it gets tricked by polarity state changes until it is attracted to the second transistor. In a ternary computer, however, you have 3 transistors you can go to next, not just 2. So now you can use the 3rd current to do so.
NOW, how the the ternary current works, is, if you are using a ternary battery for your power supply (you can use a regular un-modified battery but then you need ternary power inversion components on the computer's motherboard, I prefer a ternary source.. so...) the aluminum rod you add will have a spinning action to the current, instead of forward and back.
Looks like this:
====================>>>>> Negative to positive (1 state!)
<<======================= Positive to negative (0 state)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
==============================
VVVVVVVVVVVVVVVVVVVVVVVVVV
The above illustration is the circular or "up" direction (2 state). We say up because well, the circuitry is exposed on one side, and not the other. The current will push more "UP" than down into the silicon wafer the transistors are attached to.
So that's all 3 states. Now the aluminum attached to the battery in this case will SPIN it's current instead of going back and forth, due to being weaker than the zinc, and stronger than the positive. You COULD do this with brass but you will lessen the life of your battery, and it won't put out as much energy. Aluminum's magnetism is separated properly from the zinc and copper. Here's a diagram of how your power source will work.
http://oi60.tinypic.com/2nsrwgw.jxpxgx (change the jxpxgx to jpg)
As you can see, the diagram shows the energy flow from each terminal. Everything is attracted to the positive, but the problem is the aluminum is getting a push from the negative at the same time. This unavoidably creates a spinning action in the magnetic and electrical fields (assuming you have it hooked up and are using it). This is your 3rd current. You cannot apply this thing to normal 2-way (negative-positive) electrical components. Not unless you don't hook up the 3rd terminal to the component in question. However in a ternary computer, all the components would obviously already be there to accept the 3rd terminal and thus become capable of using it's ternary logic states.
When you apply the 3rd terminal, you can actually do many more things than just 3 state logic. You can actually make the current directional. Attaching an electromagnet to the 3 state battery will also show the capabilities of a ternary CPU by being able to manipulate a metal object, we'll just use a small machine screw for this example. The screw is magnetized to the electromagnet, and the electromagnet can move it up and down, left to right and spin the object at the same time. Using this understanding, we can obviously see that a ternary CPU would be radically advanced. Instead of requiring transistors to fire up twice to switch to a different path, you can simply back up a few transistors and put the electrical charge on a new transistor path. This saves energy, time and because of how tings are naturally multiplied, using a 3 base counting system on the ternary CPU/Computer will enable you to do math so much quicker, as proven with the RGB color scale example above.
The possibilities are ENDLESS.
精彩评论