| |
|
I think the essence of it is that it doesn't work upon individual binary digits in a straight sequence, but upon groups of binary digits.
We could imagine a rule which said "look at information coming in in groups of 4 bits; the first 4 bits you receive will tell you what to do (eg. 0001 means ADD, 0010 means SUBTRACT and so on), the second 4 bits will be a number and the third 4 bits will be another number".
We could then receive a binary string like 000100010001 and it would translate as ADD 0001 (1) to 0001 (1) and return a final string of 4 bits, 0010 (2). A presence of 0001 on the first part of the information would open up (at the hardware level) certain connections between the next two parts which would perform the ADDing operation.
Or maybe 001000010010 which would translate as SUBTRACT 0001 (1) from 0010 (2) and return 0001 (1). A presence of 0010 on the first part would open up connections which would perform the SUBTRACTing operation.
In practice it's going to be hideously more complicated but (unless I'm grossly mistaken, which is entirely possible) all computers eventually boil down to a set of hardware gates deciding what to do when they get certain groups of bits. |
|
|
