"Chip Mayan" <[log in to unmask]> wrote:

>Can anyone point me in the right direction as to how real, (Floating Point),
>numbers are stored on the HP3000?  Apparently it is not in accordance with
>IEEE standards.  We are trying to integrate information into MANMAN from an
>AS400 and want to do all the conversions on the PC side so we can minimize
>our programming on the HP side.   Thanks!
>

There are two kinds of floating point numbers on HP 3000 Series 900 systems.
There is the old HP3000 "classic" floating point used primarily by
Compatibility Mode programs still emulating the one and two decade old
architecture systems.  There is also honest-to-goodness standard IEEE floating
point in single precision (32 bits), double precision (64 bits), and to some
degree, quad precision (128 bits).  HP provides an intrinsic to convert from
one form of floating point to the other.

The single precision IEEE format, for example,  is standard.  I.e., one sign
bit, followed by 8 bits of biased binary exponent, followed by 23 bits of
fraction, with an implied leading "1."

Steve