I learned a while ago of the existence of old Soviet ternary computers and have been doing some reading about how a ternary computer would operate different from a binary one.
In a binary system, the smallest unit of data is the bit (binary digit). What would the smallest unit of data be in a ternary system?
like this
Isaac Ji Kuo
in reply to Jonathan Lamothe • • •Sensitive content
A trit.
What's really a trip is balanced ternary. It's so elegant it almost hurts.
Instead of trit values being 0, 1, and 2, the trit values are -1, 0, and 1. Each range of numbers is naturally balanced around 0, and multiplying a number by -1 is simply multiplying each digit by -1.
Jonathan Lamothe
in reply to Isaac Ji Kuo • •Jonathan Lamothe
in reply to Jonathan Lamothe • •Isaac Ji Kuo
in reply to Jonathan Lamothe • • •Sensitive content
It works the same as in decimal, except there's no need for a sign on either the mantissa or the exponent.
Base 2 floating point is a bit different because it can skip the initial 1 on the mantissa (it's always a 1, except for the edge case of zero; there needs to be exceptions for dealing with zero).
Jonathan Lamothe
in reply to Isaac Ji Kuo • •Digital Mark λ ☕️ 🕹 👽
in reply to Jonathan Lamothe • • •Sensitive content
A tit! Er, trit.
So instead of counting 1, 2, 4, 8, 16, like a normal person, it's 1, 3, 9, 27, 81.
I'm too binary-conditioned to use that even as a jokey thing. Making a VM to try it out would be hard.