P2 Taqoz V2.8 Floating Point Library

Christof Eb.

Hi,
when I had a close look at HP-85 calculator computers in the net these days, I thought, this interactive calculator mode looks very nice! It was done in BASIC then.
So, what can we do, if we want to have an interactive floating point calculator but don't want to loose all those powerful features, that are already present in Taqoz? - Do it as a RPN calculator! And of course, you can use the floating point library in your Forth programs too. Well, I know "real men do not use floating point math"....

This is an adoption of a Library by Klaus Schleisiek. - Many thanks! There is a nice description of it: http://www.complang.tuwien.ac.at/anton/euroforth/ef15/papers/schleisiek.pdf
The original source code is from: https://github.com/microCore-VHDL/microCore/blob/master/software/hfloat.fs

Unfortunately Taqoz is a non-standard Forth. So some words had to be translated. And there have been some traps hidden well. For example "2/" is unsigned in Taqoz.... It was very helpful to have the library running in GForth in parallel.

The picture shows some calculations. On Top the stack is displayed in hex, decimal, and float. The dup statements have been inserted, to preserve the old values.
Have fun!
Christof

Christof Eb.

Is it slow? Oh, yes it is: about 100 times slower than normal integer Forth and 600 times slower than optimized HUB assembler in this test.
As the floats are handled using the normal stack, SWAP and OVER act just as normal. There is much to do to add 2 floats with "f+".....

\ normal Forth version:
: fiboForth ( n -- nfibo )  \ Fibonacci Reihe, liefert letztes Ergebnis
    0 1 
    ROT 1 - FOR 
        SWAP OVER  
        + 
    NEXT
    SWAP DROP
;


: fiboFloat ( n -- nfibo )  
    0 float 1 float
    ROT 1 - FOR 
        SWAP OVER  
        f+ 
    NEXT
    SWAP DROP
    integer \ result as integer
;