=20
On Mon, 12 May 2003 08:22:50 -0400, you wrote:

>My current program requires the use of a bunch of floating point =
variables.
>Currently I was using standard 32-bit floating, but have found this =
takes up
>an enormous amount of program space.
>Is there anything else I can do different to save space.  For example, =
if I
>only want to represent numbers ranging from 0 to 9.9 (in 0.1 increments)
>etc...  Just keep
>in mind these 0 to 9.9 numbers need to be used in true 32-bit floating
>calculations.
>
>Any ideas?
>
>Thanks
>D

Do you _really_ need floating point ? Couldn't you use fixed point ?=20
remember floating point gives range, not precision - very few things that=
 embedded apps do require
enough range  to justify FP.=20

You may need to spend a little time looking at ranges of vars to =
determine where to put the fixed
point (which may vary for different variables), but you will save a lot =
of code/ data space and
time..

=46o the case of a range of 0 to 9.9, you could use an unsigned  long =
with the point between bits 21
and 20, giving a range of 0..15 with 32 bit precision. You may also be =
able to use Unsigned ints to
give 16 bit precision for some vars, which will also save a lot of space.

Depending on how input- and display- oriented things are, another =
possibility is to scale everything
up by a power of 10  to integers or longs, although using a binary point =
is usually a lot more
efficient for internal calculations, especially if the binary point is =
positioned in a multiple of 8
bits.=20

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads