From: ldo@nz.invalid   
      
   On Sat, 27 Dec 2025 18:08:38 +0200, Michael S wrote:   
      
   > Using the way you look at it (width of machine = width of its widest   
   > data register) we already have commodity general-purpose 512-bit   
   > machines for exactly ten years.   
   > But that's a wrong way to look.   
      
   In the days before byte-addressability, the “word length” of a machine   
   was a (mostly) pretty obvious thing to determine.   
      
   Byte addressability did muddy the waters somewhat. For example, the   
   original Motorola 68000 processor was widely considered to be   
   “16-bit”, even though it had 32-bit address fields and 32-bit   
   architectural registers, and the whole instruction set design was   
   clearly meant to be a cut-down 32-bit architecture. (And indeed the   
   later full-32-bit 68020 processor differed in its instruction set   
   mainly in the filling in of a few gaps.)   
      
   Whereas, “64-bit” processors were considered to be “64-bit” seemingly   
   based on their support for 64-bit addresses. Being able to operate on   
   64-bit quantities was not enough.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)