shz_scalar.h

Go to the documentation of this file.

/*! \file
 *  \brief General-purpose scalar math routines.
 *  \ingroup scalar
 *
 *  This file provides a collection of general-purpose math routines for
 *  operating on floating-point scalar values.
 *
 *  \todo
 *      - Use FP rounding modes for rounding functionality.
 *
 *  \author 2025, 2026 Falco Girgis
 *  \author 2025, 2026 Paul Cercueil
 *  \author       2026 Aleios
 *  \author       2026 jnmartin64
 *
 *  \copyright MIT License
 */
 
#ifndef SHZ_SCALAR_H
#define SHZ_SCALAR_H
 
#include <math.h>
#include <stdbool.h>
 
#include "shz_cdefs.h"
 
/*! \defgroup scalar Scalar
    \brief    Scalar functions and utilities.
 
    This API is designed around performing various scalar operations, providing
    alternatives for routines typically found within `<math.h>`.
 
    \warning
    Unlike the standard C floating-point routines, these routines are often saving
    cycles by not handling NaN and INF values as well as by not reporting rounding
    and domain errors back to the user.
*/
 
//! Floating-point epsilon used with inexact FP-based comparisons.
#define SHZ_FLT_EPSILON 0.000001f
 
SHZ_DECLS_BEGIN
 
/*! \name  Comparisons
    \brief Routines for comparing and classifying floating-point values.
    @{
*/
 
//! Returns the minimum value of two given floats.
SHZ_FORCE_INLINE float shz_fminf(float a, float b) SHZ_NOEXCEPT;
 
//! Returns the maximum value of two given floats.
SHZ_FORCE_INLINE float shz_fmaxf(float a, float b) SHZ_NOEXCEPT;
 
//! Checks for equality based on EITHER the absolute tolerance or relative tolerance, using SHZ_FLT_EPSILON.
SHZ_FORCE_INLINE bool shz_equalf(float a, float b) SHZ_NOEXCEPT;
 
//! Checks for equality based on the absolute tolerance using SHZ_FLT_EPSILON.
SHZ_FORCE_INLINE bool shz_equalf_abs(float a, float b) SHZ_NOEXCEPT;
 
//! Checks for equality based on the relative tolerance using SHZ_FLT_EPSILON.
SHZ_FORCE_INLINE bool shz_equalf_rel(float a, float b) SHZ_NOEXCEPT;
 
//! @}
 
/*! \name  Rounding
    \brief Routines for rounding and manipulating floats.
    @{
 */
 
/*! Replacement for the <math.h> routine, floorf().
 
    Returns the closest integral value to \p x, rounded down, as a float.
 
    \warning
    This routine is only valid for for the range INT32_MIN <= \p x <= INT32_MAX.
 
    \sa ceilf()
*/
SHZ_FORCE_INLINE float shz_floorf(float x) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, ceilf().
 
    Returns the closest integral value to \p x, rounded up, as a float.
 
    \warning
    This routine only returns valid values for the input range
    INT32_MIN <= \p x <= INT32_MAX.
 
    \sa floorf()
*/
SHZ_FORCE_INLINE float shz_ceilf(float x) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, roundf().
 
    Returns the value of \p x rounded to the nearest integer, as a float.
 
    \warning
    This routine only returns valid values for the input range
    -UINT32_MAX <= \p x <= UINT32_MAX.
*/
SHZ_INLINE float shz_roundf(float x) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, truncf().
 
    Returns the value of \p with its fractional component discarded.
 
    \warning
    This routine only returns valid values for the input range
    INT32_MIN <= \p x <= INT32_MAX.
*/
SHZ_FORCE_INLINE float shz_truncf(float x) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, remainderf().
 
    Returns the floating-point remainder of \p num divided by \p denom,
    rounded to the nearest integer (as a float).
 
    \warning
    This routine does not gracefully handle dividing by zero, and it is only
    valid for the input range INT32_MIN <= \p num / \p denom <= INT32_MAX.
 
    \sa shz_fmodf(), shz_remquof(), shz_truncf().
*/
SHZ_FORCE_INLINE float shz_remainderf(float num, float denom) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, fmodf().
 
    Returns the floating-point remainder of \p num divided by \p denom,
    rounded towards zero.
 
    \warning
    This routine does not gracefully handle dividing by zero.
 
    \sa shz_remainderf()
*/
SHZ_FORCE_INLINE float shz_fmodf(float num, float denom) SHZ_NOEXCEPT;
 
/*! (Sorta) Replacement for the <math.h> routine, remquof().
 
    Returns the floating-point remainder of \p num divided by \p denom,
    rounded to the nearest integer (as a float). \p quot is set equal to
    the quotient which is used as part of the calculation.
 
    \note
    \p quot is returning as a `float` rather than an `int` as with standard C.
    Simply cast to an `int` manually afterwards if that is the desired behavior.
 
    \warning
    This routine does not gracefully handle dividing by zero.
 
    \sa shz_remainderf()
*/
SHZ_FORCE_INLINE float shz_remquof(float num, float denom, float* quot) SHZ_NOEXCEPT;
 
//! @}
 
/*! \name  Mapping
    \brief Routines for mapping a number to another range.
    @{
*/
 
//! Clamps a floating-point value by the given \p min and \p max values.
SHZ_FORCE_INLINE float shz_clampf(float value, float min, float max) SHZ_NOEXCEPT;
 
//! Maps a value within the given range \p from to \p to, to be within the range of `0.0f += 1.0f`.
SHZ_FORCE_INLINE float shz_normalizef(float current, float from, float to) SHZ_NOEXCEPT;
 
//! Maps a value within the given range \p from to \p to, to be within the range of `0.0f + 1.0f` more quickly, provided \p to - \p from is a positive difference.
SHZ_FORCE_INLINE float shz_normalizef_fsrra(float current, float from, float to) SHZ_NOEXCEPT;
 
//! Maps a value within the given range \p inputStart to \p inputEnd, to be within the range of \p outputStart to \p outputEnd.
SHZ_FORCE_INLINE float shz_remapf(float value, float inputStart, float inputEnd, float outputStart, float outputEnd) SHZ_NOEXCEPT;
 
//! Maps a value within the given range \p inputStart to \p inputEnd, to be within the range of \p outputStart to \p outputEnd more quickly, provided the \p outputEnd - \p outputStart is a positive difference.
SHZ_FORCE_INLINE float shz_remapf_fsrra(float value, float inputStart, float inputEnd, float outputStart, float outputEnd) SHZ_NOEXCEPT;
 
//! Wraps the given \p value back to be within the range of \p min to \p max.
SHZ_FORCE_INLINE float shz_wrapf(float value, float min, float max) SHZ_NOEXCEPT;
 
//! Wraps the given \p value back to be within the range of \p min to \p max more quickly, provided \p max - \p min is a positive difference.
SHZ_FORCE_INLINE float shz_wrapf_fsrra(float value, float min, float max) SHZ_NOEXCEPT;
 
 
//! Returns the fractional part of \p x, equivalent to GLSL `fract()`.
SHZ_FORCE_INLINE float shz_fractf(float x) SHZ_NOEXCEPT;
 
//! Returns -1.0f if \p x < 0, 0.0f if \p x == 0, or 1.0f if \p x > 0.
SHZ_FORCE_INLINE float shz_signf(float x) SHZ_NOEXCEPT;
 
//! Clamps \p x to the range [0.0f, 1.0f].
SHZ_FORCE_INLINE float shz_saturatef(float x) SHZ_NOEXCEPT;
 
//! @}
 
/*! \name  Miscellaneous
 *  \brief Assorted routines implementing other fp operations.
 *  @{
 */
 
/*! Replacement for the <math.h> routine, fabsf().
 
    Returns the absolute value of \p x.
*/
SHZ_FORCE_INLINE float shz_fabsf(float x) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, copysignf().
 
    Returns the value of \p x with the sign of \p y.
*/
SHZ_FORCE_INLINE float shz_copysignf(float x, float y) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, fmaf().
 
    Returns \p a * \p b + \p c, performing an FP multiply + accumulate operation.
*/
SHZ_FORCE_INLINE float shz_fmaf(float a, float b, float c) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, fdimf(),
 
    Returns the positive difference between \p x and \p y or zero if
    y >= x.
 
    \warning
    Unlike fdimf(), this routine does not handle INF and NAN values.
*/
SHZ_FORCE_INLINE float shz_fdimf(float x, float y) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, hypotf().
 
    Returns the hypoteneuse of the right triangle with the given legs.
 
    \warning
    Unlike hypotf(), this routine has no error or overflow handling.
*/
SHZ_FORCE_INLINE float shz_hypotf(float x, float y) SHZ_NOEXCEPT;
 
/*! Replacement for the <math.h> routine, cbrtf().
 
    Returns the cubed root of \p x, using a fast approximation.
*/
SHZ_FORCE_INLINE float shz_cbrtf(float x) SHZ_NOEXCEPT;
 
//! Returns a value that is linearly interpolated between \p a and \p b by the given ratio, \p t.
SHZ_FORCE_INLINE float shz_lerpf(float a, float b, float t) SHZ_NOEXCEPT;
 
//! Returns a value that is barycentrically interpolated between \p a, \p b, and \p c using the given barycentric coordinates, \p u and \p v.
SHZ_FORCE_INLINE float shz_barycentric_lerpf(float a, float b, float c, float u, float v) SHZ_NOEXCEPT;
 
//! Uses the quadratic formula with the given coefficients to solve for the two roots, returning `true` if any real roots exist, and `false` if the roots are only imaginary.
SHZ_FORCE_INLINE bool shz_quadratic_roots(float a, float b, float c, float* root1, float* root2) SHZ_NOEXCEPT;
 
//! Returns a random floating-point number between `0.0f` and `1.0f`, using and updating the given seed.
SHZ_FORCE_INLINE float shz_randf(int* seed) SHZ_NOEXCEPT;
 
//! Returns a random floating-point number between \p min and \p max, using and updating the given seed.
SHZ_FORCE_INLINE float shz_randf_range(int* seed, float min, float max) SHZ_NOEXCEPT;
 
//! returns 0.0f if x < edge, otherwise 1.0f
SHZ_FORCE_INLINE float shz_stepf(float x, float edge) SHZ_NOEXCEPT;
 
//! Returns 0.0f at/below edge0, 1.0f at/above edge1, smoothly varying in-between. edge0 must be less than edge1 or result is undefined.
SHZ_FORCE_INLINE float shz_smoothstepf(float x, float edge0, float edge1) SHZ_NOEXCEPT;
 
//! Returns 0.0f at/below edge0, 1.0f at/above edge1, smoothly varying in-between. Accepts inverse edges.
SHZ_FORCE_INLINE float shz_smoothstepf_safe(float x, float edge0, float edge1) SHZ_NOEXCEPT;
 
//! @}
 
/*! \name  FSRRA
 *  \brief Routines built around fast reciprocal square root instruction.
 *  @{
 */
 
//! Calculates 1.0f/sqrtf( \p x), using a fast approximation.
SHZ_FORCE_INLINE float shz_inv_sqrtf_fsrra(float x) SHZ_NOEXCEPT;
 
//! Calculates 1.0f/sqrtf( \p x ), using a fast approximation, while safely protecting against division-by-zero.
SHZ_FORCE_INLINE float shz_inv_sqrtf(float x) SHZ_NOEXCEPT;
 
//! Returns the fast approximate square root of the given value, \p x.
SHZ_FORCE_INLINE float shz_sqrtf_fsrra(float x) SHZ_NOEXCEPT;
 
//! Returns the fast approximate square root of the given value, \p x, safely returning `0.0f` is \p x == `0.0f`.
SHZ_FORCE_INLINE float shz_sqrtf(float x) SHZ_NOEXCEPT;
 
//! Takes the inverse of \p x using a very fast approximation, returning a positive result.
SHZ_FORCE_INLINE float shz_invf_fsrra(float x) SHZ_NOEXCEPT;
 
//! Takes the inverse of \p x using a slighty faster approximation than doing a full division, safely handling negative values.
SHZ_FORCE_INLINE float shz_invf(float x) SHZ_NOEXCEPT;
 
//! Divides \p num by \p denom using a very fast approximation, which requires \p denom be a positive value.
SHZ_FORCE_INLINE float shz_divf_fsrra(float num, float denom) SHZ_NOEXCEPT;
 
//! Divides \p num by \p denom using a slightly faster approximation, allowing \p denom to be negative.
SHZ_FORCE_INLINE float shz_divf(float num, float denom) SHZ_NOEXCEPT;
 
//! @}
 
/*! \name  FIPR
 *  \brief Routines built around fast 4D dot product.
 *  \todo  Generalize these with a macro that lets you choose vector FP regs.
 *  @{
 */
 
//! Takes two sets of 3D vectors as 3 floats and calculates their dot product using an approximation.
SHZ_FORCE_INLINE float shz_dot6f(float x1, float y1, float z1,
                                 float x2, float y2, float z2) SHZ_NOEXCEPT;
 
//! Takes two sets of 4D vectors as 4 floats and calculates their dot product using an approximation.
SHZ_FORCE_INLINE float shz_dot8f(float x1, float y1, float z1, float w1,
                                 float x2, float y2, float z2, float w2) SHZ_NOEXCEPT;
 
//! Takes a 3D vector as 3 floats and calculates its squared magnitude using a fast approximation.
SHZ_FORCE_INLINE float shz_mag_sqr3f(float x, float y, float z) SHZ_NOEXCEPT;
 
//! Takes a 4D vector as 4 floats and calculates its squared magnitude using a fast approximation.
SHZ_FORCE_INLINE float shz_mag_sqr4f(float x, float y, float z, float w) SHZ_NOEXCEPT;
 
//! @}
 
/*! \name  Transcendental
    \brief Fast approximations for non-trig transcendental functions.
    @{
*/
 
//! Fast approximation for raising 2 to a floating-point power.
SHZ_FORCE_INLINE float shz_pow2f(float p) SHZ_NOEXCEPT;
 
//! Fast approximation for C's powf().
SHZ_FORCE_INLINE float shz_powf(float x, float p) SHZ_NOEXCEPT;
 
//! Fast approximation of POSIX's pow10f().
SHZ_FORCE_INLINE float shz_pow10f(float x) SHZ_NOEXCEPT;
 
//! Fast approximation for C's log2f().
SHZ_FORCE_INLINE float shz_log2f(float x) SHZ_NOEXCEPT;
 
//! Fast approximation for C's logf().
SHZ_FORCE_INLINE float shz_logf(float x) SHZ_NOEXCEPT;
 
//! Fast approximation for C's log10f().
SHZ_FORCE_INLINE float shz_log10f(float x) SHZ_NOEXCEPT;
 
//! Fast approximation for C's expf().
SHZ_FORCE_INLINE float shz_expf(float p) SHZ_NOEXCEPT;
 
//! @}
 
#include "inline/shz_scalar.inl.h"
 
SHZ_DECLS_END
 
#endif // SHZ_INTRIN_H