mirror/tracy
1
0
Fork 0
mirror of https://github.com/wolfpld/tracy.git synced 2024-11-10 02:31:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

Apply IM_NORMALIZE2F_OVER_ZERO optimization.

Browse Source
This commit is contained in:
Bartosz Taudul 2021-05-01 17:11:59 +02:00
parent 0e6a25fdd8
commit 630615c6c2
No known key found for this signature in database
GPG Key ID: B7FE2008B7575DF3
2 changed files with 12 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
imgui/imgui_draw.cpp
View File

@ -694,7 +694,7 @@ void ImDrawList::PrimQuadUV(const ImVec2& a, const ImVec2& b, const ImVec2& c, c
// On AddPolyline() and AddConvexPolyFilled() we intentionally avoid using ImVec2 and superfluous function calls to optimize debug/non-inlined builds.
// Those macros expects l-values.
#define IM_NORMALIZE2F_OVER_ZERO(VX,VY) do { float d2 = VX*VX + VY*VY; if (d2 > 0.0f) { float inv_len = 1.0f / ImSqrt(d2); VX *= inv_len; VY *= inv_len; } } while (0)
#define IM_NORMALIZE2F_OVER_ZERO(VX,VY) do { float d2 = VX*VX + VY*VY; if (d2 > 0.0f) { float inv_len = ImRsqrt(d2); VX *= inv_len; VY *= inv_len; } } while (0)
#define IM_FIXNORMAL2F(VX,VY) do { float d2 = VX*VX + VY*VY; if (d2 < 0.5f) d2 = 0.5f; float inv_lensq = 1.0f / d2; VX *= inv_lensq; VY *= inv_lensq; } while (0)
// TODO: Thickness anti-aliased lines cap are missing their AA fringe.

12
imgui/imgui_internal.h
View File

@ -51,6 +51,10 @@ Index of this file:
#include <math.h> // sqrtf, fabsf, fmodf, powf, floorf, ceilf, cosf, sinf
#include <limits.h> // INT_MIN, INT_MAX
#if defined __SSE__ || defined __x86_64__ || defined _M_X64
#include <immintrin.h>
#endif
// Visual Studio warnings
#ifdef _MSC_VER
#pragma warning (push)
@ -386,6 +390,12 @@ 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); }
#if defined __SSE__ || defined __x86_64__ || defined _M_X64
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)
@ -406,7 +416,7 @@ 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 ImVec2 ImFloor(const ImVec2& v) { return ImVec2((float)(int)(v.x), (float)(int)(v.y)); }
static inline int ImModPositive(int a, int b) { return (a + b) % b; }