Skybuck's Universal Code Version 3

Bad drug interaction?

Luhan

Wow a post from Starbuck that is less than a 100 lines

Rehab must be working

martin

Cool indeed. You have just frustrated the plans of all those people who were busy patenting ways to cut communication channel capacity in half.

John

Dead from the neck, up.

-- Service to my country? Been there, Done that, and I've got my DD214 to prove it. Member of DAV #85.

Michael A. Terrell Central Florida

Nah, just showing that Shannon was an optimist and PT Barnum a pessimist.

If the "no more bits" information was not included, the channel could handle twice the amount of information.

BTW: If you used the right base, only 3 states would be needed for your method.

True it's trade off.

What good would that do for a binary computer and theory ? ;)

Bye, Skybuck.

Then why don't you simply add a '0' after a string of, say, five '1's and use six '1's as a marker? That way you only add one bit for every 32, on average. If you're smart, you could even use this information to extract a clock (you know there is a transition at least every six bits). You could call it "Skybucks Data Link Control", or something.

What does any of Skybuck's flaming ideas have to with computers or theory?

This is not my area of expertise, but I have seen posts about this elsewhere, and this appears to be an attempt to compress information in a bit stream. When there are states other than 0 and 1, it is no longer binary, and it becomes an analog system, which may then be expanded to its digital binary equivalent.

For some types of data, such as text, a variable length frame (as I would call it) might be more efficient. I think Morse Code is something like that, where the most common letters E and T are represented by a single DIT or DAH. However, to represent this in binary, a frame separator would be two zeroes, and a DIT would be a 1 and a DAH would be 11. So, an E would take three bits, T would take four bits, and a zero (five DAHs) would be 17 bits. The average would be about 11 bits for about 40 unique characters, although probably 63 could be represented. 8 bit asynchronous ASCII can represent 256 characters with a fixed frame size of 10 bits, so it appears to be more efficient.

Synchronous data, requiring a separate line or another analog state on a single line, could be used for a more efficient variable frame scheme. A sync pulse would start each successive frame, which could be shorter for more frequently used codes. It would probably only save one bit for random binary data, and you lose that bit with the sync pulse.

I don't think there is any new clever way to pack more data in a binary stream.

Paul

Yes, you are trading away 50% of the bandwidth of the channel to do what can be done with less than 40%.

The computations would still be done in a simple binary manner. With 3 states, you can carry all the information needed to do what you want with

Because that involves adding an extra bit.

With Skybuck's Universal Code Version 3 the CPU registers would remain the same bit size.

Nice isn't it ;)

Bye, Skybuck.

Cool idea, to use morse code as compression for text. I wonder if huffman or morse code is more efficient :)

Anyway thanks for posting this because morse code might be more easy to implement than huffman... and maybe morse code is faster to encode/decode... who knows what properties it might have :)

You lost me on this one ;) But thats ok :)

The question is how to encode/decode/describe the length of the data ? ;)

That's what Universal Codes are about.

Bye, Skybuck.

Probably :)

However, lol, using more states can give certain properties which 3 states won't have ;)

Bye, Skybuck.

When did you stop sniffing glue?

In article , Skybuck wrote: [...]

No, all the suggestion here have been for you to leave out a bit. With your code, every bit of information requires two bits to encode. This is a huge waste of bits.

I'm not sure if we run out of ones or zeros first but you are taking twice your share.

In article , Skybuck wrote: [....]

Yes, it has the property of wasting bits, being slow and requiring extra hardware.

Maybe, maybe not.

But you missed the advantage property that's for sure.

See other thread for advantage property ;)

Bye, Skybuck.

Nope,

You look at it the wrong way.

There are only 2 bits, zero and one.

The code is for hardware, the hardware simply uses different voltages to encode/decode the bits.

Voltagesbits ;)

Bye, Skybuck.

In article , Skybuck wrote: [...]

log2(NumberOfVoltages) = bits. If you have 4 levels, you have 2 bits.