fx

Header-only fixed-point arithmetic C implementation for embedded systems without a dedicated FPU. Leverages compiler constant folding to initialize fixed-point numbers like floating-point. Only requires stdint.h.

Features

• 32-bit (Q23.8, adjustable) fixed-point creation and resolution
• Common arithmetic functions
• Ceil and floor functions

Example

#include "fx.h"

fx32 f = fx32_make(1, 500);             // Create an approximation of 1.5f
f = fx32_mul(f, fx32_make(2, 0));       // Multiply it by 2.0f
int i = fx32_resolve(f)                 // Resolve `f` to 3

Macros

Define these before including fx.h to adjust its behavior.

Macro	Usage	Example
FX_NO_INLINE	Remove compiler inline hints	#define FX_NO_INLINE
FX_FAST	Avoid mitigating overflow by up-casting during multiplication/division (saves processor cycles)	#define FX_FAST
FX32_SHIFT_BITS	Define the amount of bits used for fx32 fractional part (default 8 bits)	#define FX32_SHIFT_BITS 8
FX32_QUANTIZE_LIMIT	Define the upper limit of fx32 fractional part's quantized range (default 1000, avoid exceeding 2^16)	#define FX32_QUANTIZE_LIMIT 1000