The DDP 24 FloatingPoint Format, and Others

Finally, I found out what the floating-point format of the DDP-24 was; it was in the last manual I downloaded (from Al Kossow's site, or rather one of its mirrors)...

it was similar to the weird formats common on other 24-bit computers (I discuss the rationale on my page at

which I believe has to do with the fact that these machines, in general, did multiplication in hardware, but not floating-point) but it had one ADDITIONAL bit of weirdness.

As it used sign-magnitude representation, it placed the sign of the exponent in the sign bit of the second word, not contiguous with the exponent at the *end* of the word.

I buttume, but am not certain, that the software floating-point format on the SDS 920-930-940 computers was identical to the hardware floating-point format on the 9300; perhaps downloading the FORTRAN manuals for these machines will explain this, or perhaps I can peruse certain other things I've already downloaded.

Still mysterious is the floating-point format of the CDC 924. As this machine performed 24 x 24 = 48 bit multiplication in conventional integer form, with no attempt to make the result work as a proper fraction without shifting instead, I suspect this one's complement machine used a *conventional* floating-point format, like that of the 1604, but I cannot be certain.

John Savard