home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   comp.arch      Apparently more than just beeps & boops      131,241 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 129,458 of 131,241   
   Waldek Hebisch to John Savard   
   Re: Changes May Be Coming   
   21 Aug 25 16:08:43   
   
   From: antispam@fricas.org   
      
   John Savard  wrote:   
   > I found and corrected some errors in the opcode space allocated to headers.   
   > Some bit patterns were overlapping with instructions.   
   > This happened as I was preparing to change the ISA of Concertina II. I   
   > thought that the 14-bit short instructions weren't really all that useful,   
   > and I would be better off by using the four-bit pattern it consumed to   
   > indicate a header indicating a block consisting of seven 36-bit   
   > instructions.   
   >   
   > But that seemed like just making a third basic instruction set format...   
   > with all three of them being different ways of doing the same thing.   
   > Something else suggested itself to me instead.   
   >   
   > If I took the various methods to make more compact instructions, and   
   > combined them all, going all out, I could make memory-reference   
   > instructions that were only 22 bits long. That would leave enough room to   
   > allow them to be _predicated_ without a header!   
   >   
   > But that had its own problems. Predication and break bits... were what was   
   > normally associated with a VLIW instruction set. And even if those didn't   
   > need headers, I still didn't see a way to avoid them for pseudo-   
   > immediates. This might reduce the need for headers and block structure,   
   > and allow the scheme of headers to be simplified, but it wouldn't mean I   
   > could do without it. So would this radical change _really_ make things   
   > better?   
   >   
   > So I'm getting to that point I dreaded - the one where instead of settling   
   > on _this_ iteration of Concertina II, I go back to going around in   
   > circles. Will I find a satisfying genuine improvement?   
      
   Your messages suggest that you are building your instruction encoding   
   by hand.  There are rather old algorithms like Huffman compression   
   and optimal search tree.  If you define instructions and add weight   
   factors (importance or expected frequency) appropriate algorithm   
   shoul be able to pack your instructions as densly as possible   
   and guarantee that there are no overlap.   
   --   
                                 Waldek Hebisch   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca