thacks42 / fixed_point_math

a templated header-only fixed point math library for C++

fixed point math library

this repository contains a templated, header-only fixed point math library.

Features

supported features so far:

• arbitrary integer types and number of fractional bits
• all basic arithmetic operations plus square root
• boolean comparisons
• construction from user defined literals
• conversion from/to strings
• utility functions to access the internal bit representation
• compiles to "the correct assembly" on both gcc and clang using O1 and up.

Example usage

using namespace fixed_point;
using fp32_16 = fixed<int32_t, 16>;
fp32_16 a = 2.173_fixp_t;
fp32_16 b = 5.183_fixp_t;
auto c = a + b;
auto d = make_fixed<int32_t, 16>(2);
auto e = c * d - a;
debug_print(e); //prints: "binary:            1100.1000100111111011 as double:      12.538986206055"

Headers

fixed_point_type.hpp:

contains the basic type definitions as well as anything related to construction and basic conversions

fixed_point_math.hpp:

contains the arithmetic operations

fixed_point_print.hpp:

contains helpers to print fixed point values. Makes use of doubles as well as including fmt/core.h and iostream. Do not include this on a target that doesn't have these libraries or floating point support! This header is meant for debugging purposes.

fixed_point_float_conversions.hpp:

contains functions to convert from/to floats. Do not include this if you don't have floating point support on your target!