Possibly another way to physically represent data other than just bits?

Bobdude161 · Jul 16, 2009

This is a theory based off what I have learned. Must have an understanding of binary to continue.

In our hard drives we use on and off to represent ones and zeros that make up all our data. Many things that are stored on there are ASCII letters and numbers, MP3s and JPEGs. ASCII uses 8 bits, MP3s use 128 bits and JPEGs use 24 bits. (Correct me if I'm wrong here). Each format is divisible by four. When we want to represent the letter "a" in binary we use "01100001", because we are limited to two options 1 and 0. Since a lot of our data formats are more or equal to 4 bits most times (right?) we can find other ways to represent data than just ones and zeros. Instead of 1 and 0, how about 0-7? Confused yet? Lemme explain.

Instead of using binary, we could use quartets. So instead of using "0110 0001" to represent "a" we could use "6 1". 0110 = 6 and 0001 = 1. Now how would 0-7 be represented on a physical data storage device? Not with on and off but with 8 different possible states. 8 states for each number 0-7. Those 8 states could be represented by 8 different pit depths for optical drives or 8 different magnetic states for HDs. Having 1 quartet state instead 4 binary states could increase storage capacity quite a bit.

I'm not sure how CPUs could use this kind of addressing. Maybe using different levels of electrical power for those 8 states. It's all kind of hurting my head right now.

Does this make sense? Could this really be used? Or is it more complicated than generalizing 4 bits into one quartet? Thoughts?

dmr727 · Jul 16, 2009

Google both the octal and hexadecimal (base-8 and base-16) numbering systems. Since just about all our digital circuits are transistor based (i.e. on or off), binary tends to be the flavor of choice.

r.j.s · Jul 16, 2009

While technically possible. I don't see it happening. Everything for a computer would have to be redone, and what would be the real benefit?

chrmjenkins · Jul 16, 2009

The format you're suggesting is actually octal format (or base 8). To convert a binary number to any base that is a power of two, you simply section off the bits x at a time, where x is the power of two of the base you're converting to, and translate those blocks into numbers.

As for optical drives, depth of pits wouldn't work. The pits are used to scatter the light, depriving a detector of picking up a reflection. A pit is simply going to scatter the light until you make it so shallow that it behaves like a land. To do anything to a finer degree would become prohibitively complex.

As for CPUs, this wouldn't work either. The advantage of a binary system is that you high two clearly defined voltages, Vd and Ground. Having only two states allows for some variance, or more properly defined as noise, without losing the actual data. To try and represent more data approaches analog circuits, which, while carrying more information in a single signal, must also permit some amount of imprecision.

Quantum computers, which use qubits for information storage, actually have more than two states per qubit (it's actually four). Although I'll admit I don't now a great deal about these systems, they are very interesting, and they will be the standard in 50 or so years' time.

yg17 · Jul 16, 2009

Pretty much what everyone else said.

Everything can be represented as on or off, whether it's magnetic like a hard drive, a pit like a CD, electricity like a connector (CPU, USB, SATA, DVI, etc) or light like fiber. That allows some variance. For example, the laser at the end of the fiber may have to be a certain brightness, but some variance is allowed as long as the receiver at the other end can tell if it's on or off. Using different brightnesses to represent multiple values would require extreme precision which would not be cost prohibitive. Same with electricity levels, pit depth, etc. Using 1s and 0s allows devices to be less than perfect.

sushi · Jul 17, 2009

There have been computers that do use a system other than binary.

One historical exception are the few balanced and non-balance ternary computers.

r.j.s · Jul 17, 2009

Here's a concept biological computer: http://news.bbc.co.uk/2/hi/science/nature/358822.stm

Signal-11 · Jul 17, 2009

sushi said:
There have been computers that do use a system other than binary.

One historical exception are the few balanced and non-balance ternary computers.

Yeah, but if you're going to go that far, may as well call Babbage's difference engine a base 10 computer.

m85476585 · Jul 17, 2009

Maybe the magnetic bits in a hard drive could be set to 8 different angles, but I think it would be much more expensive and complicated, and it would be easier to lose data if one bit moves a little.

CPUs and memory probably have to remain binary.

sushi · Jul 18, 2009

Signal-11 said:
Yeah, but if you're going to go that far, may as well call Babbage's difference engine a base 10 computer.

Well, there have been electronic computers based upon the ternary system.

Search

Search

Possibly another way to physically represent data other than just bits?

Bobdude161

macrumors 65816

dmr727

macrumors G4

r.j.s

Moderator emeritus

chrmjenkins

macrumors 603

yg17

macrumors Pentium

sushi

Moderator emeritus

r.j.s

Moderator emeritus

Signal-11

macrumors 65816

m85476585

macrumors 65816

sushi

Moderator emeritus

Our Staff