Fixed point math library

[viki2000]

The concept is here: https://en.wikipedia.org/wiki/Fixed-point_arithmetic
I would like to generate a sine using a fixed point math library as Microchip has in "libq.h" with _Q15sinPI on page 232 next file:
http://ww1.microchip.com/downloads/en/DeviceDoc/50001456J.pdf
http://ww1.microchip.com/downloads/en//softwarelibrary/fixed%20point%20math%20library/50001456j.pdf

The idea is to have the argument of the function and the returned value in the range -32768 to 32767, no (real) floating numbers.
The existing sin() function from "math.h" described in PCD manual page 190, has as argument of the function and the returned value floating numbers:
https://www.ccsinfo.com/downloads/PCDReferenceManual.pdf

The way how I do it now is implicit, typecasting, by assigning the float variable value to an signed int16 variable:

[temtronic]

You don't say which PIC you're using or the resolution you need but I'd think a lookup table for 0-90* would/could be very fast and efficient.
It would depend upon what you need the numbers for....
Decades ago there was an article about 'magic sine waves' by Don Lancaster.

Jay

[viki2000]

I use PIC24HJ64GP202 or generally a PIC24 on 16 bit MCU.
I want to generate the sine wave using 1000 up to 10.000 points.
At 10K points the lookup table is not a nice option, actually even for 1000 points is big., so I decided to try a trigonometric function as sin().
I use 16bit for one DAC sample as resolution.