> Forget Newton's algorithm; too slow and inaccurate for my taste.  Have a
> look at the two attachments for square root for primitive PICs.  This is
> original code, by the way, and has been thoroughly tested.  I have a
> version
> for higher end PIC's as well but somehow I've managed to lose it.  I'll
> keep
> looking for it.

Hi Bruce,

You may wish to check out this square root routine for the 12 and 14-bit
cores:

http://www.dattalo.com/technical/software/pic/picsqrt.html


As far as Olin's original equation, the square root is only one component.
There is also the division and multiplication. Another confounding factor
is that the equation is a function of two variables. If speed is the
ultimate requirement, then there are ways to implement the equation
efficiently. First, one must specify the accuracy requirements. For
example, is it okay that T1 has only 4-bits of dynamic range or does it
need 8? Once this is known, then the appropriate approach is more obvious.
(E.g. brute force like it is already implemented, look up tables, chordic,
series expansion, or a combination of all of these)


I suspect there are other reasons besides this equation to use the dsPIC
family.

Scott
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist