Drezil/imgui
1
0
Fork 0
mirror of https://github.com/Drezil/imgui.git synced 2025-07-04 12:08:47 +02:00
Code Issues Packages Projects Releases Wiki Activity

Add and use SSE-enabled ImRsqrt() in place of 1.0f / ImSqrt(). (#4091)

Browse Source 
Squashed 3 commits.
This commit is contained in:
Bartosz Taudul
2021-05-06 17:25:40 +02:00
committed by ocornut
parent 84545dbe6f
commit 4c9f0cec27
3 changed files with 15 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
imgui_internal.h
View File

@ -51,6 +51,12 @@ Index of this file:
#include <math.h> // sqrtf, fabsf, fmodf, powf, floorf, ceilf, cosf, sinf
#include <limits.h> // INT_MIN, INT_MAX
// Enable SSE intrinsics if available
#if defined __SSE__ || defined __x86_64__ || defined _M_X64
#define IMGUI_ENABLE_SSE
#include <immintrin.h>
#endif
// Visual Studio warnings
#ifdef _MSC_VER
#pragma warning (push)
@ -390,6 +396,12 @@ static inline float ImAbs(float x) { return fabsf(x); }
static inline double ImAbs(double x) { return fabs(x); }
static inline float ImSign(float x) { return (x < 0.0f) ? -1.0f : ((x > 0.0f) ? 1.0f : 0.0f); } // Sign operator - returns -1, 0 or 1 based on sign of argument
static inline double ImSign(double x) { return (x < 0.0) ? -1.0 : ((x > 0.0) ? 1.0 : 0.0); }
#ifdef IMGUI_ENABLE_SSE
static inline float ImRsqrt(float x) { return _mm_cvtss_f32(_mm_rsqrt_ss(_mm_set_ss(x))); }
#else
static inline float ImRsqrt(float x) { return 1.0f / sqrtf(x); }
#endif
static inline double ImRsqrt(double x) { return 1.0 / sqrt(x); }
#endif
// - ImMin/ImMax/ImClamp/ImLerp/ImSwap are used by widgets which support variety of types: signed/unsigned int/long long float/double
// (Exceptionally using templates here but we could also redefine them for those types)
@ -410,7 +422,7 @@ static inline ImVec4 ImLerp(const ImVec4& a, const ImVec4& b, float t)
static inline float ImSaturate(float f) { return (f < 0.0f) ? 0.0f : (f > 1.0f) ? 1.0f : f; }
static inline float ImLengthSqr(const ImVec2& lhs) { return (lhs.x * lhs.x) + (lhs.y * lhs.y); }
static inline float ImLengthSqr(const ImVec4& lhs) { return (lhs.x * lhs.x) + (lhs.y * lhs.y) + (lhs.z * lhs.z) + (lhs.w * lhs.w); }
static inline float ImInvLength(const ImVec2& lhs, float fail_value) { float d = (lhs.x * lhs.x) + (lhs.y * lhs.y); if (d > 0.0f) return 1.0f / ImSqrt(d); return fail_value; }
static inline float ImInvLength(const ImVec2& lhs, float fail_value) { float d = (lhs.x * lhs.x) + (lhs.y * lhs.y); if (d > 0.0f) return ImRsqrt(d); return fail_value; }
static inline float ImFloor(float f) { return (float)(int)(f); }
static inline float ImFloorSigned(float f) { return (float)((f >= 0 || (int)f == f) ? (int)f : (int)f - 1); } // Decent replacement for floorf()
static inline ImVec2 ImFloor(const ImVec2& v) { return ImVec2((float)(int)(v.x), (float)(int)(v.y)); }