Fursure.
What makes you think I missed what you claim is an advantage? There is no real advantage. Its an illusion.
Fursure.
What makes you think I missed what you claim is an advantage? There is no real advantage. Its an illusion.
In a binary sysyetm of course by definition.
He's talking about a base 4 method.
Graham
If you follow Skybuck's idea to its logical conclusion you end up with analogue.
Graham
Ok, log4*Number of voltages) = qits.
It's still 2 bits (or one qit). If he dropped one of the levels he'd have a...
You only add one bit every thirty two, on average.
No, because they all have to be twice as big as they otherwise need be.
Morse code IS a Huffman code.
...and why you're so poor at designing them.
On 16 Jul 2006 09:50:01 -0700, "Skybuck" Gave us:
You're f****ng 100% lost.
On Sat, 22 Jul 2006 03:04:36 +0100, Eeyore Gave us:
Save space? Power? Speed? Asynchronous clocking, and other little niceties like CML (Current Mode Logic).
What's currently the best analog schema? QAM 256 or such? What one can pack into a carrier comes down to modulation schema, right?
See QRSL...
Scairy stuff.
John
Quartinary equivelent to having 2 bits per digit or per bit time. He, however is only getting one bit of information from most of the digits. A lot of modems used quartinary modulation methods to pack more information into a give time. This isn't what he is doing.
In article , krw wrote: [...]
On this point you are wrong. This is not a case of adding one more bit. It is a case of leaving 31 out. There is no "extra" bit just one that couldn't be left out.
... or the mad house.
No, to express a symbol with 5 '1's in a row, one must insert an _extra_ '0'. The chances of 5 or more '1's in a row is 1/32.
It is. One always adds the _extra_ bit after five consecutive '1's. If that's not adding one bit, I don't know what is.
Nope, but I'd like what you're smoking.
I think you misunderstood. Under the Skybuck system, you need to carry two bits for every bit of data. Taking his system as the starting point every system that has been suggested in this group has been a suggestion for what bits could be left out without loosing the ability to handle variable length data.
Considering your example above, under Skybucks system, you would need 10 bits to transmit the data but you've only needed 6 bits in the worst case. So you've been able to leave 4 out. So your improvement over Skybuck's system can't leave out one bit.
Yes, I see now that we're talking across each other.
I was responding to Skybucks response to my suggestion of "Skybucks Data Link Control" (SDLC). SDLC's '0' insertion *is* an extra bit every 32ish.
But that's not what this branch of the thread was about.
In article , krw wrote: [... me ...]
Yes I guess we were. Now back to our normal programming
Let's be clear about something.
Huffman is not a universal code !
Bye, Skybuck.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required