On Wed, Aug 07, 2002 at 06:27:20PM -0400, Ross Patterson wrote:
> On Wednesday 07 August 2002 05:24 pm, David Laight wrote:
> > I wrote code for one system that
> > did arithemtic on decimal numbers (up to 10 digits) with
> > a sign 'overpunch'.
> 
> Sounds like we're both former IBM mainframe programmers. :-)

No - wrong guess.

ASCII art of layout of each 10 byte instruction:

  Byte    0   1   2   3   4   5   6   7   8   9
	+-------+-------+---+---+---+-------+---+
  Bit 8	| 0   0 |  PAa  | 0 |EXM| 0 |  PBa  | 0 |
	+-------+-------+---+---+---+-------+---+
  Bit 7	|       Fb      |AC |  IA   |  IB   |BC |
	+---------------+---+-------+-------+---+
  Bit 6	|  ignored - normally inverse of bit 7  |
	+---+-----------+---+---+-----------+---+
  Bit 5	|IDA|    PAb    |Fa |IDB|    PBb    |EIX|
	+---+-----------+---+---+-----------+---+
  Bit 4 |   |               |   |               |
	+   +               +   +               +
  Bit 3 |   |               |   |               |
	+LA +      A        +LB +      B        +
  Bit 2 |   |               |   |               |
	+   +               +   +               +
  Bit 1 |   |               |   |               |
	+---+---------------+---+---------------+

Opcodes are (typically) dyadic, length up to 10 bytes (each),
address upto 320000 (PAa and PAb are binary extensions to
the decimal A).

Absolutely no binary arithmetic at all (binary AND, OR and XOR),
single instructions to do disk read/write and multiply/divide.

	David

-- 
David Laight: david@l8s.co.uk