[inforoots] first use of bits to store programmes?

Michael Albaugh m.e.albaugh at gmail.com
Thu Nov 23 15:10:41 PST 2006


On 11/22/06, Niels Ole Finnemann <finneman at imv.au.dk> wrote:
> Help: I suppose there are people on this list who can help me with the
> following question:
> It's known that Konrad Zuse used binary notation for numbers in his
> calculating machines, that von Neumann developed the idea of a stored
> program, and that "Claude Shannon identified the bit as the
> fundamental unit of data and, coincidentally, the basic unit of
> computation". (comp hist timeline) but now I want to know who was the
> first to use the binary alphabet for representation (and execution) of
> programmes - and not simply data?

Well, Zuse's programs were also stored in a binary form,
as punched (or not) holes in recycled film-stock. His was what
would later be called a "Harvard Architecture" machine, with
separate stores for program and data (film and mechanical
latches, respectively). Some things to note on that:
1) "Harvard" machines were so called in distinction to
   "Princeton" (or IAS) machines, which had a unified store
   for program and data.
2) Modern use of "Harvard Architecture" is often to refer to
   machines with separate caches for Instruction and data,
   but fed from a unified store.
3) Some machines even today truly separate instructions
    from data (e.g 8051 or PIC)
4) The machine described by Von Neuman (see Ed Thelen's
    post for provenance) did store program and data in the
    same physical store, but the use of a given word was
    determined at program-load time. If a word was loaded
    with an instruction, it _stayed_ that instruction. Storing
    to an instruction modified only the address part (for
    array references, before the invention of the B-Box)
    "executing" a data word acted as a "Load Immediate"
    of its value to the accumulator. "Self Modifying code"
    in the modern sense would not have been possible.

(Yeah, I know "too much information :-)

Does Tim want to chime in one whether the Analytic Engine
micro-store is "binary", or could a Cam have multiple levels?

   Mike



More information about the inforoots mailing list