Loading ...
Sorry, an error occurred while loading the content.

Octal

Expand Messages
  • Bill Degnan
    ... I did not live through the times, but I can see where using octal to program and test computers would have had its advantage. It s easier to do octal in
    Message 1 of 7 , Jun 23, 2008
    • 0 Attachment
      -------- Original Message --------
      > From: "William Donzelli" <wdonzelli@...>
      > Sent: Monday, June 23, 2008 5:53 PM
      >
      > > There was a recent MARCH post about the "accident" of a byte being 8
      > > bits. Part of the issue of processor "word" size over the decades, was
      > > in dealing with arithmetic.
      >
      > During the development of the IBM System/360 during the early 1960s,
      > when the world started to shift to the 8 bit standard, dealing with
      > arithmetic was hugely influential. Early in the project's engineering
      > cycle, IBM studied data usage, and found that an overwhelming amount
      > of data in the world at the time was numeric. Storage was also
      > insanely expensive. 8 bit bytes were shown to be a very good
      > compromise between storage and processor efficiency.
      >
      > Those firms that still were using 6 and 9 bit multiples generally saw
      > the error of their early design choices, and most eventually adopted
      > architectures in multiples of 8 bits - even stubborn DEC and CDC
      > eventually changed. Univac never really did the change, and their
      > architecture is one of the very last still being produced that has to
      > do all sorts of hairy conversions to talk to an 8 bit world.
      >
      > --
      > Will
      >

      I did not live through the times, but I can see where using octal to program and test computers would have had its advantage. It's easier to do octal in your head.
      bd
    • Bob Applegate
      That s exactly what a lot of people said. I never liked octal and had no problem with hex, but a lot of the early guys took the you can pry octal from my
      Message 2 of 7 , Jun 23, 2008
      • 0 Attachment
        That's exactly what a lot of people said.  I never liked octal and had no problem with
        hex, but a lot of the early guys took the "you can pry octal from my dead hands" sort
        of approach. 

        Personally, I think hex math is pretty easy; it's whatever you get used to.  People not
        comfortable with octal made all sorts of mistakes even though it was "easier". 

        Learning to work in ANY new number base is difficult until you get used to it.  After
        all, a lot of people don't get the common T-shirt that reads:

            "There are 10 kinds of people in this world...
            "Those who get binary and those who don't."

        Boy, is that one difficult to explain to a non-computer user, :-)

        Bob

        I did not live through the times, but I can see where using octal to program and test computers would have had its advantage. It's easier to do octal in your head. 
        bd 


      • Mike Loewen
        ... I m with you, Bob. I got very comfortable with hex writing assembly code on a Z80, then later on was exposed to octal notation on the SAGE. I still
        Message 3 of 7 , Jun 23, 2008
        • 0 Attachment
          On Mon, 23 Jun 2008, Bob Applegate wrote:

          > Personally, I think hex math is pretty easy; it's whatever you get used
          > to. People not comfortable with octal made all sorts of mistakes even
          > though it was "easier".

          I'm with you, Bob. I got very comfortable with hex writing assembly
          code on a Z80, then later on was exposed to octal notation on the SAGE. I
          still prefer hex. Funny how Unix file permissions are still specified in
          octal.


          Mike Loewen mloewen@...
          Old Technology http://sturgeon.css.psu.edu/~mloewen/Oldtech/
        • Dan Roganti
          Mike Loewen wrote: On Mon, 23 Jun 2008, Bob Applegate wrote: Personally, I think hex math is pretty easy; it s whatever you get used to. People not comfortable
          Message 4 of 7 , Jun 23, 2008
          • 0 Attachment


            Mike Loewen wrote:
            On Mon, 23 Jun 2008, Bob Applegate wrote:
            
              
            Personally, I think hex math is pretty easy; it's whatever you get used 
            to. People not comfortable with octal made all sorts of mistakes even 
            though it was "easier".
                
                I'm with you, Bob.  I got very comfortable with hex writing assembly 
            code on a Z80, then later on was exposed to octal notation on the SAGE.  I 
            still prefer hex.  Funny how Unix file permissions are still specified in 
            octal.
              
            It was funny how the Altair 8800 is an 8bitter an yet the console was arranged in octal. As the Intellec before that, had a generic arrangement on it's front console, which implied (sort of) using Hex. Other micros, shortly thereafter, mostly used hex. I always wondered what rationale the folks at MITS had in their minds. Then I built the Altair 680b kit which did have their front console (and coding) arranged using hexadecimal. Then later the Altair 8800b was made but still arranged in octal.

            Hexadecimal numbering convention was available since the 50's by IBM but the letters took sometime to get standardized with A-F.

            =Dan
            [ Pittsburgh 250th --- http://www2.applegate.org/~ragooman/   ]
            

          • Herb Johnson
            ... Look at the 8080 instructions in binary. There are eight references to registers or memory - three bits. instructions which move or compare values use two
            Message 5 of 7 , Jun 24, 2008
            • 0 Attachment
              Dan wrote:

              >It was funny how the Altair 8800 is an 8bitter an yet
              >the console was arranged in octal. As the Intellec before that...
              > I always wondered what rationale the folks at MITS
              > had in their minds.

              Look at the 8080 instructions in binary. There are eight references to
              registers or memory - three bits. instructions which move or compare
              values use two sets of those three bits, leaving two bit for
              instruction types. An OCTAL representation of 8080 op codes is the
              most consistent format.

              Herb Johnson
              retrotechnology.com
            • Dan Roganti
              Herb Johnson wrote: Look at the 8080 instructions in binary. There are eight references to registers or memory - three bits. instructions which move or compare
              Message 6 of 7 , Jun 25, 2008
              • 0 Attachment


                Herb Johnson wrote:
                Look at the 8080 instructions in binary. There are eight references to
                registers or memory - three bits. instructions which move or compare
                values use two sets of those three bits, leaving two bit for
                instruction types. An OCTAL representation of 8080 op codes is the
                most consistent format.
                  

                I did, but only the register opcodes follow this convention, that's only about 20% of all the opcodes in there. The others each have various bit fields which don't fit into the octal format.

                =Dan

              • Herb Johnson
                ... Note: This is an arcane discussion today, but hex versus octal was an ACTIVE discussion in the mid-late 1970 s. I was there, and part of it, and there
                Message 7 of 7 , Jun 27, 2008
                • 0 Attachment
                  Herb Johnson wrote:
                  >> Look at the 8080 instructions in binary. There are eight
                  >> references to registers or memory - three bits. instructions
                  >> which move or compare
                  >> values use two sets of those three bits, leaving
                  >> two bit for instruction types. An OCTAL representation
                  >> of 8080 op codes is the most consistent format.

                  Dan said:

                  > I did, but only the register opcodes follow this convention,
                  > that's only about 20% of all the opcodes in there. The others
                  > each have various bit fields which don't fit into the octal
                  > format.

                  Note: This is an arcane discussion today, but "hex versus octal" was
                  an ACTIVE discussion in the mid-late 1970's. I was there, and part of
                  it, and there were reasons for it.

                  It is true that only 8080 register-explicit instructions follow the
                  EXACT pattern I described. But if you inspect the other instructions
                  carefully, they follow a consistent scheme of binary patterns, like
                  the register scheme. I mentioned the "register" convention to suggest
                  that pattern. I did not want to post a lecture on 8080 op codes.

                  I might have said that in the 8080, there are consistent instruction
                  bit fields, shown from left to right, as bit 6&7, 3-5, and 0-2. (7 is
                  the most signifigant bit.) For example, the "jump" instructions all
                  have bits 6&7 = "11", and almost all have bits 2-0 ="010". Call
                  instructions, bits 6&7 = "11", bits 2-0 = "100". The exceptions in
                  both cases are the unconditional jump and call. There are similar
                  patterns in other instruction groups.

                  But, it's a perceptual issue as well, and different people "see"
                  patterns or not in different ways. This should not become a personal
                  discussion.

                  My point was that these patterns are clearer in OCTAL representations
                  of op codes, than in HEXidecimal. Intel groups these instructions in
                  BINARY charts in their data books, but organized by instruction type.
                  I'm using an Intel 1978 8085 data book right now. These patterns
                  reflect the underlying digital binary logic which decodes the
                  instructions; there are reasons for such patterns.

                  The fact it WAS a debate, can be found in articles from microcomputer
                  magazines of the period. If some today find this discussion arcane, or
                  boring, or trivial; it was an issue at a time when people DID program
                  in binary and think in binary. Front panels weren't for show, they
                  were used. They were used in an era when many people were writing
                  their own assemblers, and when the only digital storage was paper tape
                  or audio cassette - or pencil and paper.

                  Herb Johnson
                  retrotechnology.com
                Your message has been successfully submitted and would be delivered to recipients shortly.