Dear U S Software Customer:

Congratulations on your purchase of the F/DPAC Floating Point Library.

We suggest that you consider the disk containing the F/DPAC source files
to be a "master" disk.  Please copy the source files to another disk to
produce working copies and save the master disk as a backup.

The files comprising the F/DPAC delivery package are:

    FPOPNS.S      Single precision floating point operations
    FPFNCS.S      Single precision floating point functions
    FPCNVT.S      Single precision conversion routines

    DPOPNS.S      Double precision floating point operations
    DPFNCS.S      Double precision floating point functions
    DPCNVT.S      Double precision conversion routines

    DPPAC.H       DPAC/FPAC Header File
    DPTEST.C      Testbed/demonstration program
    DPTEST.OUT    Executable module of DPTEST

    DPOPNS.TST    Automatic tests for DP floating point operations
    DPFNCS.TST    Automatic tests for DP floating point functions
    DPCNVT.TST    Automatic tests for DP conversion routines
    FPOPNS.TST    Automatic tests for SP floating point operations
    FPFNCS.TST    Automatic tests for SP floating point functions
    FPCNVT.TST    Automatic tests for SP conversion routines

The floating point demonstration program allows the user to exercise
the features of the DPAC/FPAC library and to experiment with them.  The
program simulates a stack calculator with a four deep (maximum) stack,
quite similar in fact to HP's hand held calculators.  Two stacks are
maintained, one for single precision floating point numbers and the other
for double precision floating point numbers.  Additionally, two sets of
registers are maintained for the G and S commands.

Binary operations are done between the top two items on the stack, with
one item being consumed in performing the binary operation.  Functions
and unary operations are performed on the top stack item.  Automatic
testing may be initiated by using the B command.

The testbed is designed to operate on an Intel 80960KB board within an
IBM PC compatible platform.  The testbed may be initiated in this
environment by entering the following:

                sdm_kb              start 80960KB board
                reset               reset 80960KB board
                ~d dptest.out       download DPAC testbed
                go 0x10000000       start DPAC testbed

The user is prompted for a command.  After the command is entered; the
specified command action is taken and a display of the stack contents is
produced in hex representation and a decimal interpretation of that hex.
The chart below shows the F/DPAC routine being exercised, the testbed
command, and the command action.


F/DPAC     CMD   ACTIONS               F/DPAC    CMD   ACTIONS
---------  ----  ------------------    --------  ----  ------------------
f/dascbin  fpn   X = fpn                         Q     QUIT 80960 TestBed
           Hhex  X = hex                         B     Batch Test Mode
           M     toggle SP/DP                    =     Batch compare X & Y
f/dpadd    +     X = Y + X                       ?     X = compare X & Y
f/dpsub    -     X = Y - X             f/dpsub   _     X = X - Y
f/dpmul    *     X = Y * X             f/dpdiv   /     X = Y / X
f/dpdiv    \     X = Y / X                       Ffpn  X = fpn
                                       f/dpsqrt  FR    X = sqrt(X)
f/daint    A     X = floor(X)          f/dpflt   Fint  X = int
f/dint     I     X = int(X)            f/dpsin   FS    X = sin(X)
f/dfix     Z     X = fix(X)            f/dpcos   FC    X = cos(X)
f/dbinasc  C     X -> ASCII            f/dptan   FT    X = tan(X)
sptodp     D     X = dp(X)             f/dpatn   FA    X = atn(X)
dptosp     S     X = sp(X)             f/dpexp   FE    X = X to e pow
           R     roll X>Y>Z>T>X        f/dpxtoi  F^int X = X to int pow
           Gint  X = f/dreg[int]       f/dpln    FN    X = log e  (X)
           Sint  f/dreg[int] = X       f/dllog   FL    X = log 10 (X)

Notes:
-----------------------------------------------------------------------
1]  fpn is a floating point number
2]  hex is a hexidecimal number
3]  int is an integer
4]  X is the top of the stack followed by Y,Z,T
5]  Trigonometric functions deal with radian quantities
6]  The dynamic range of single precision IEEE floating
      point numbers is  +/- 1.2E-38 to 3.4E+38
7]  The dynamic range of double precision IEEE floating
      point numbers is  +/- 2.225E-308 to 1.797E+308
8]  An invalid operation (zero divided by zero, log of a negative
      number, etc...) will produce a Not-a-Number result (NaN)
9]  Infinity is represented; it is printed as +/- INF


Good programming!