Re: [Sdcc-user] Floating point support

[D H. <ill...@sl...>]

Hi All!

I enjoy reading this list, even if I don't have much time for uC coding 
these days...

I will agree with George's position -- don't do hard things unless 
there's a compelling reason.  That's also a good reason for using SDCC 
instead of assembly.

I did run into a compelling reason to do "floating point" multiplication 
using 8-bit registers on a PIC14.  At first I brought in some float 
library code, but there was no room when the project was finished, so I 
switched over.  It certainly helps if you have a decimal->hex 
calculator!  I even limited my mult routines to one: 16x16 unsigned 
mult, to save space.  It was an interesting exercise, and much easier to 
go through the second time when I decided I needed an extra decimal 
digit of resolution!

I found it was great for calibrations and linearization formulae for 
sensor data -- a fixed application with known input and output ranges.

There seems to be plenty of resources on the net in this area, but I'll 
include snippets of code, if there is interest.

Doug