[CCC DEV] Floating point on Arduino...

Matt Jadud jadudm at gmail.com
Tue Jun 8 20:30:07 BST 2010


On Tue, Jun 8, 2010 at 3:21 PM, cljacobsen at gmail.com
<cljacobsen at gmail.com> wrote:
> I forget under what conditions we are able to use the floating point
> emulation library... however... interpreted on that machine I think
> you might as well attach a slide rule interface to the board?

Yeah.

The problem/challenge is this:

The UAV group is looking at implementing filtering and other functions
on the readings from the gyros and accelerometers. Many of these
algorithms are written by converting everything into units like volts,
or radians/second (or degrees/second, etc.)... and while we could do
everything in integer maths, we would loose something. Perhaps a lot
of something.

This is a problem that some students have bumped up against before,
but it wasn't a very big deal. People working on the Arduino are
accustomed to being able to use floating point numbers... do we
convert everything to integer mathematics? Another approach would be
to pass everything in and out of C as INT32s, and do the floating
point operations there with support (say) from avr-libc.

Thoughts?

M




More information about the developers mailing list