To use, get a compiler that supports C++14 (gcc and clang are supported, Visual Studio will currently not compile the code, and other compilers have not been tested), and
#include <FixedPoint/FixedPoint.hpp>
Then declare some variables with an integer width, a fractional width, and optionally an unsigned storage type, which defaults to 32-bit ints. The integer and fractional width need to add up to one less than a multiple of the size of the storage type.
The variables
FixedPoint<5, 26> a;
FixedPoint<5, 26, uint32_t> b;
FixedPoint<5, 26, uint16_t> c;
have the same integer precision of 5 bits, and the same fractional precision
of 26 bits. a and b both use 32-bit numbers for storage, while c uses 16-bit
numbers.
Arithmetics work as expected:
FixedPoint<5, 26> a, b, c;
a = 5.;
b = 6.;
c = a/b;
printf("%g / %g = %g\n", double(a), double(b), double(c));
will print
5 / 6 = 0.833333
Other than the standard arithmetic (+, -, *, /, % and their
assignment equivalents) and comparison operators, the following functions
are defined as well:
std::sin
std::cos
std::atan2
std::abs
std::exp
std::pow
std::log
For the trigonometric functions, the results are at the moment only tested with an integral width of 3 and will probably return nonsensical results with smaller widths.
For more examples, look at the provided tests in FixedPointTests.cpp