Sign magnitude, one's complement, two's complement

[Chuck Guzis]

On 08/22/2015 02:23 PM, Sean Conner wrote:
>
> For my own morbid curiosity, and because it came up on another
> mailing list I'm on [1], what machines commercially avaialble were
> sign magnitude and one's complement?  Every machine I've encountered
> was two's complement (okay, IEEE 754 [2] is a sign magnitude format
> but I'm talking about integer implementations here, not floating
> point).  I've only found reference to one sign magnitude computer
> (the IBM 7090, release in 1959) and a few one's complement machines
> (mostly the PDP series from DEC).
>
> Where there others?  And honestly, are there any machines that use
> anything other than two's complement today?

There were, as noted, several ones complement systems.

Every decimal computer that I can think of, or ones, at least that have 
a decimal instruction vocabulary are sign-magnitude.  That is, a 
negative one is represented as -1 in some form, not in its nine's or 
ten's complement.

For another pure binary sign-magnitude system, consider one of the first 
minicomputers (although the term hadn't been then invented)--the Packard 
Bell PB 250.  I'm pretty certain there were several others.

Floating point can engender some interesting representations.  Consider 
the exponent field on the aforementioned CDC 6000 series.  It's a 
"biased by 2000 octal) system--and the assumed binary point of the 
mantissa is to the right of the LSB.  So, 2000 0000 0000 0000 0001 octal 
= 1 exactly.

--Chuck