Re: [avr-libc-dev] Pow Function in avr8

[Georg-Johann Lay]

Thomas, George schrieb:

> > > Also adding to the points mentioned above, is it required to use
> > > the logrithms for getting this done as we explicitly say the
> > > second argument of pow should be an integer ?

This is rather an optimization for the compiler and than map pow to a 
powi something.

IIRC GCC performs that so the right agenda is to verify that GCC 
performs the transformation for a primary target like x86 and then find 
out why that it is not performed on avr.

Maybe a missing TARGET_C99_FUNCTIONS macro is the reason or because 
avr-gcc has non-ISO double.

> My suggestion was handling the cases where power required is a whole
> number or an integer separately so that accuracy for them can be
> improved rather than going by the exponent method.

If you know that the exponent is integral you may want to have a look at 
GCC's __builtin_powi* functions.

http://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html

You asked for a base of 2, maybe ldexp() fits your use case.

Handling integral exponents separately will increase the code size 
because the exponent must be checked at run time and extra code must be 
executed in that case.

Johann