--User changes

-Improve performance in the following variations: cpow2, dc_cube, julia3d, julia3dz, julian2, log_db, nblur, npolar, waffle, wavesn, xtrb.

--Code changes
 -Rand range now uses multiply + shift rather than modulo.

Source/Ember/Ember.h

@@ -1078,11 +1078,11 @@ public:
 			};
 
 			if (rand.Rand() & 1)
-				sym = symDistrib[rand.Rand() % Vlen(symDistrib)];
+				sym = symDistrib[rand.Rand(Vlen(symDistrib))];
 			else if (rand.Rand() & 31)
-				sym = intmax_t(rand.Rand() % 13) - 6;
+				sym = intmax_t(rand.Rand(13)) - 6;
 			else
-				sym = intmax_t(rand.Rand() % 51) - 25;
+				sym = intmax_t(rand.Rand(51)) - 25;
 		}
 
 		if (sym == 1 || sym == 0)

Source/Ember/Isaac.h

@@ -148,9 +148,9 @@ public:
 	/// </summary>
 	/// <param name="upper">A value one greater than the maximum value that will be returned</param>
 	/// <returns>A value between 0 and the value passed in minus 1</returns>
-	inline T Rand(T upper)
+	inline T Rand(size_t upper)
 	{
-		return (upper == 0) ? Rand() : Rand() % upper;
+		return (upper == 0) ? Rand() : T(((size_t)Rand() * upper) >> 32);
 	}
 
 	/// <summary>
@@ -158,7 +158,7 @@ public:
 	/// </summary>
 	/// <param name="upper">A value one greater than the maximum value that will be returned</param>
 	/// <returns>A value between 0 and the value passed in minus 1</returns>
-	static inline T LockedRand(T upper)
+	static inline T LockedRand(size_t upper)
 	{
 		rlg l(*s_CS.get());
 		T t = GlobalRand->Rand(upper);
@@ -420,14 +420,14 @@ protected:
 		{
 #ifndef __ISAAC64
 			RngStep((a << 13), a, b, mm, m, m2, r, x, y);
-			RngStep((a >> 6) , a, b, mm, m, m2, r, x, y);
+			RngStep((a >> 6), a, b, mm, m, m2, r, x, y);
-			RngStep((a << 2) , a, b, mm, m, m2, r, x, y);
+			RngStep((a << 2), a, b, mm, m, m2, r, x, y);
 			RngStep((a >> 16), a, b, mm, m, m2, r, x, y);
 #else   // __ISAAC64
 			RngStep(~(a ^ (a << 21)), a, b, mm, m, m2, r, x, y);
-			RngStep(  a ^ (a >> 5)  , a, b, mm, m, m2, r, x, y);
+			RngStep(  a ^ (a >> 5), a, b, mm, m, m2, r, x, y);
-			RngStep(  a ^ (a << 12) , a, b, mm, m, m2, r, x, y);
+			RngStep(  a ^ (a << 12), a, b, mm, m, m2, r, x, y);
-			RngStep(  a ^ (a >> 33) , a, b, mm, m, m2, r, x, y);
+			RngStep(  a ^ (a >> 33), a, b, mm, m, m2, r, x, y);
 #endif  // __ISAAC64
 		}
 
@@ -437,14 +437,14 @@ protected:
 		{
 #ifndef __ISAAC64
 			RngStep((a << 13), a, b, mm, m, m2, r, x, y);
-			RngStep((a >> 6) , a, b, mm, m, m2, r, x, y);
+			RngStep((a >> 6), a, b, mm, m, m2, r, x, y);
-			RngStep((a << 2) , a, b, mm, m, m2, r, x, y);
+			RngStep((a << 2), a, b, mm, m, m2, r, x, y);
 			RngStep((a >> 16), a, b, mm, m, m2, r, x, y);
 #else   // __ISAAC64
 			RngStep(~(a ^ (a << 21)), a, b, mm, m, m2, r, x, y);
-			RngStep(  a ^ (a >> 5)  , a, b, mm, m, m2, r, x, y);
+			RngStep(  a ^ (a >> 5), a, b, mm, m, m2, r, x, y);
-			RngStep(  a ^ (a << 12) , a, b, mm, m, m2, r, x, y);
+			RngStep(  a ^ (a << 12), a, b, mm, m, m2, r, x, y);
-			RngStep(  a ^ (a >> 33) , a, b, mm, m, m2, r, x, y);
+			RngStep(  a ^ (a >> 33), a, b, mm, m, m2, r, x, y);
 #endif  // __ISAAC64
 		}
 

Source/Ember/PaletteList.cpp

@@ -269,7 +269,7 @@ Palette<T>* PaletteList<T>::GetRandomPalette()
 	while (attempts < Size() * 10)
 	{
 		auto p = s_Palettes.begin();
-		auto paletteFileIndex = QTIsaac<ISAAC_SIZE, ISAAC_INT>::LockedRand() % Size();
+		auto paletteFileIndex = QTIsaac<ISAAC_SIZE, ISAAC_INT>::LockedRand(Size());
 		size_t i = 0;
 
 		//Move p forward i elements.
@@ -281,7 +281,7 @@ Palette<T>* PaletteList<T>::GetRandomPalette()
 
 		if (i < Size())
 		{
-			size_t paletteIndex = QTIsaac<ISAAC_SIZE, ISAAC_INT>::LockedRand() % p->second.size();
+			size_t paletteIndex = QTIsaac<ISAAC_SIZE, ISAAC_INT>::LockedRand(p->second.size());
 
 			if (paletteIndex < p->second.size() && !p->second[paletteIndex].IsEmpty())
 				return &p->second[paletteIndex];

Source/Ember/SheepTools.h

@@ -254,7 +254,7 @@ public:
 			//Generate a 2-xform random.
 			Random(mutation, useVars, sym, 2, maxVars);
 			//Which xform to mutate?
-			modXform = m_Rand.Rand() % ember.TotalXformCount();
+			modXform = m_Rand.Rand(ember.TotalXformCount());
 			auto xform1 = ember.GetTotalXform(modXform);
 			auto xform2 = mutation.GetTotalXform(0);
 			os << "mutate xform " << modXform << " coefs";
@@ -280,7 +280,7 @@ public:
 		else if (mode == eMutateMode::MUTATE_POST_XFORMS)
 		{
 			bool same = (m_Rand.Rand() & 3) > 0;//25% chance of using the same post for all of them.
-			size_t b = 1 + m_Rand.Rand() % 6;
+			size_t b = 1 + m_Rand.Rand(6);
 			os << "mutate post xforms " << b << (same ? " same" : "");
 
 			for (size_t i = 0; i < ember.TotalXformCount(); i++)
@@ -387,7 +387,7 @@ public:
 		}
 		else if (mode == eMutateMode::MUTATE_DELETE_XFORM)
 		{
-			size_t nx = m_Rand.Rand() % ember.TotalXformCount();
+			size_t nx = m_Rand.Rand(ember.TotalXformCount());
 			os << "mutate delete xform " << nx;
 
 			if (ember.TotalXformCount() > 1)
@@ -626,7 +626,7 @@ public:
 		}
 		else
 		{
-			ember.AddXforms(xformDistrib[m_Rand.Rand() % Vlen(xformDistrib)]);
+			ember.AddXforms(xformDistrib[m_Rand.Rand(Vlen(xformDistrib))]);
 			addfinal = m_Rand.Frand01<T>() < T(0.15);//Add a final xform 15% of the time.
 
 			if (addfinal)
@@ -639,7 +639,7 @@ public:
 
 		//If useVars is empty, randomly choose one to use or decide to use multiple.
 		if (useVars.empty())
-			var = m_Rand.RandBit() ? m_Rand.Rand() % varCount : -1;
+			var = m_Rand.RandBit() ? m_Rand.Rand(varCount) : -1;
 		else
 			var = -2;
 
@@ -689,7 +689,7 @@ public:
 				else if (multid && var == -1)
 				{
 					if (xform->TotalVariationCount() < maxVars)
-						xform->AddVariation(m_VariationList->GetVariation(m_Rand.Rand() % varCount)->Copy());//Choose a random var for this xform.
+						xform->AddVariation(m_VariationList->GetVariation(m_Rand.Rand(varCount))->Copy());//Choose a random var for this xform.
 				}
 				else
 				{
@@ -722,7 +722,7 @@ public:
 								if (var != -2)
 								{
 									//Pick a random variation and use a random weight from 0-1.
-									Variation<T>* v = m_VariationList->GetVariationCopy(static_cast<size_t>(m_Rand.Rand() % varCount), m_Rand.Frand<T>(T(0.001), 1));
+									Variation<T>* v = m_VariationList->GetVariationCopy(static_cast<size_t>(m_Rand.Rand(varCount)), m_Rand.Frand<T>(T(0.001), 1));
 
 									if (v && !xform->AddVariation(v))
 										delete v;//It already existed and therefore was not added.
@@ -730,7 +730,7 @@ public:
 								else
 								{
 									//Pick a random variation from the suppled IDs and use a random weight from 0-1.
-									Variation<T>* v = m_VariationList->GetVariationCopy(useVars[m_Rand.Rand() % useVars.size()], m_Rand.Frand<T>(T(0.001), 1));
+									Variation<T>* v = m_VariationList->GetVariationCopy(useVars[m_Rand.Rand(useVars.size())], m_Rand.Frand<T>(T(0.001), 1));
 
 									if (v && !xform->AddVariation(v))
 										delete v;
@@ -760,12 +760,12 @@ public:
 						if (var != -2)
 						{
 							//Pick a random variation and use a random weight from 0-1.
-							xform->AddVariation(m_VariationList->GetVariationCopy(static_cast<size_t>(m_Rand.Rand() % varCount), m_Rand.Frand<T>(T(0.001), 1)));
+							xform->AddVariation(m_VariationList->GetVariationCopy(static_cast<size_t>(m_Rand.Rand(varCount)), m_Rand.Frand<T>(T(0.001), 1)));
 						}
 						else
 						{
 							//Pick a random variation from the suppled IDs and use a random weight from 0-1.
-							xform->AddVariation(m_VariationList->GetVariationCopy(useVars[m_Rand.Rand() % useVars.size()], m_Rand.Frand<T>(T(0.001), 1)));
+							xform->AddVariation(m_VariationList->GetVariationCopy(useVars[m_Rand.Rand(useVars.size())], m_Rand.Frand<T>(T(0.001), 1)));
 						}
 					}
 				}
@@ -779,7 +779,7 @@ public:
 		}
 
 		//Randomly add symmetry (but not if we've already added a final xform).
-		if (sym || (!(m_Rand.Rand() % 4) && !addfinal))
+		if (sym || (!(m_Rand.Rand(4)) && !addfinal))
 			ember.AddSymmetry(sym, m_Rand);
 		else
 			ember.m_Symmetry = 0;
@@ -932,7 +932,7 @@ public:
 
 		while (ntries++ < 100)
 		{
-			size_t i = m_Rand.Rand() % ember.TotalXformCount();
+			size_t i = m_Rand.Rand(ember.TotalXformCount());
 
 			if (i != excluded)
 			{

Source/Ember/Variations03.h

@@ -1777,7 +1777,7 @@ public:
 		T temp = (helper.m_PrecalcAtanyx + M_2PI * rand.Rand(uint(m_AbsN))) / m_N;
 		helper.Out.x = r * std::cos(temp);
 		helper.Out.y = r * std::sin(temp);
-		helper.Out.z = r * helper.In.z / (helper.m_PrecalcSqrtSumSquares * m_AbsN);
+		helper.Out.z = r * helper.In.z / Zeps(helper.m_PrecalcSqrtSumSquares * m_AbsN);
 	}
 
 	virtual string OpenCLString() const override
@@ -1796,11 +1796,16 @@ public:
 		   << "\n"
 		   << "\t\tvOut.x = r * cos(temp);\n"
 		   << "\t\tvOut.y = r * sin(temp);\n"
-		   << "\t\tvOut.z = r * vIn.z / (precalcSqrtSumSquares * " << absn << ");\n"
+		   << "\t\tvOut.z = r * vIn.z / Zeps(precalcSqrtSumSquares * " << absn << ");\n"
 		   << "\t}\n";
 		return ss.str();
 	}
 
+	virtual vector<string> OpenCLGlobalFuncNames() const override
+	{
+		return vector<string> { "Zeps" };
+	}
+
 	virtual void Precalc() override
 	{
 		m_AbsN = std::abs(m_N);

Source/Ember/Variations06.h

@@ -859,7 +859,7 @@ private:
 		if (params.ExactCalc == 1)
 			angXY = rand.Frand01<T>() * M_2PI;
 		else
-			angXY = (std::atan(params.ArcTan1 * (rand.Frand01<T>() - T(0.5))) / params.ArcTan2 + T(0.5) + T(rand.Rand() % glm::uint(params.NumEdges))) * params.MidAngle;
+			angXY = (std::atan(params.ArcTan1 * (rand.Frand01<T>() - T(0.5))) / params.ArcTan2 + T(0.5) + T(rand.Rand(glm::uint(params.NumEdges)))) * params.MidAngle;
 
 		sincos(angXY, &params.X, &params.Y);
 		angMem = angXY;

Source/EmberCL/EmberCLFunctions.h

@@ -174,7 +174,7 @@ static const char* RandFunctionString =
 	"\n"
 	"inline uint MwcNextRange(uint2* s, uint val)\n"
 	"{\n"
-	"	return (val == 0) ? MwcNext(s) : (MwcNext(s) % val);\n"
+	"	return (val == 0) ? MwcNext(s) : (uint)(((ulong)MwcNext(s) * (ulong)val) >> 32);\n"
 	"}\n"
 	"\n"
 	"inline real_t MwcNext01(uint2* s)\n"

Source/EmberGenome/EmberGenome.cpp

@@ -625,7 +625,7 @@ bool EmberGenome(int argc, _TCHAR* argv[], EmberOptions& opt)
 			if (opt.CloneAction() != "")
 				os << " " << opt.CloneAction();
 
-			selp0 = embers[rand.Rand() % embers.size()];
+			selp0 = embers[rand.Rand(embers.size())];
 			save = selp0;
 			aselp0 = &selp0;
 			aselp1 = nullptr;
@@ -643,7 +643,7 @@ bool EmberGenome(int argc, _TCHAR* argv[], EmberOptions& opt)
 
 				if (doMutate)
 				{
-					selp0 = embers[rand.Rand() % embers.size()];
+					selp0 = embers[rand.Rand(embers.size())];
 					orig = selp0;
 					aselp0 = &selp0;
 					aselp1 = nullptr;
@@ -685,8 +685,8 @@ bool EmberGenome(int argc, _TCHAR* argv[], EmberOptions& opt)
 				}
 				else if (doCross0)
 				{
-					i0 = rand.Rand() % embers.size();
+					i0 = rand.Rand(embers.size());
-					i1 = rand.Rand() % embers2.size();
+					i1 = rand.Rand(embers2.size());
 					selp0 = embers[i0];
 					selp1 = embers2[i1];
 					aselp0 = &selp0;

Source/EmberTester/EmberTester.cpp

@@ -2240,42 +2240,79 @@ int _tmain(int argc, _TCHAR* argv[])
 {
 	//int i;
 	bool b = true;
+	size_t times = 1'000'000'001;
+	QTIsaac<ISAAC_SIZE, ISAAC_INT> rand;
+	std::vector<unsigned int> vec(16, 0);
 	Timing t(4);
-	/*  vector<Ember<float>> fv;
+	/*
-	    vector<Ember<double>> dv;
+		for (size_t i = 1; i < times; i++)
-	    list<Ember<float>> fl;
+		{
-	    list<Ember<double>> dl;
+			auto res = rand.Rand() % i;
-	    int w = 1000, h = 1000;
+			vec[res & 15]++;
-	    string filename = ".\\testexr.exr";
+		}
-	    vector<Rgba> pixels;
-	    pixels.resize(w * h);
 
-	    for (auto& pix : pixels)
+		t.Toc("rand mod");
-	    {
-	    pix.r = 1.0;
-	    pix.b = 0.0;
-	    pix.a = 1.0;
-	    //pix.r = std::numeric_limits<float>::max();
-	    }
 
-	    writeRgba1(filename.c_str(), pixels.data(), w, h);
+		for (auto& it : vec)
-	    TestFuncs();
+		{
-	    string line = "title=\"cj_aerie\" smooth=no", delim = " =\"";
+			cout << it / (double)times << endl;
-	    auto vec = Split(line, delim, true);
+			it = 0;
+		}
 
-	    for (auto& s : vec) cout << s << endl;
+		cout << "\n\n";
+		t.Tic();
 
-	    line = "index=0 color=2177354", delim = " =";
+		for (size_t i = 1; i < times; i++)
-	    vec = Split(line, delim, true);
+		{
+			//auto res = (uint)(((size_t)rand.Rand() * i) >> 32);
+			auto res = rand.Rand(i);
+			vec[res & 15]++;
+		}
 
-	    for (auto& s : vec) cout << s << endl;
+		t.Toc("rand mult shift");
+
+		for (auto& it : vec)
+		{
+			cout << it / (double)times << endl;
+			it = 0;
+		}
+
+		return 1;
+		  vector<Ember<float>> fv;
+		    vector<Ember<double>> dv;
+		    list<Ember<float>> fl;
+		    list<Ember<double>> dl;
+		    int w = 1000, h = 1000;
+		    string filename = ".\\testexr.exr";
+		    vector<Rgba> pixels;
+		    pixels.resize(w * h);
+
+		    for (auto& pix : pixels)
+		    {
+		    pix.r = 1.0;
+		    pix.b = 0.0;
+		    pix.a = 1.0;
+		    //pix.r = std::numeric_limits<float>::max();
+		    }
+
+		    writeRgba1(filename.c_str(), pixels.data(), w, h);
+		    TestFuncs();
+		    string line = "title=\"cj_aerie\" smooth=no", delim = " =\"";
+		    auto vec = Split(line, delim, true);
+
+		    for (auto& s : vec) cout << s << endl;
+
+		    line = "index=0 color=2177354", delim = " =";
+		    vec = Split(line, delim, true);
+
+		    for (auto& s : vec) cout << s << endl;
 
 
-	    EmberContainerTester<float>::TestEmberContainer(fv);
+		    EmberContainerTester<float>::TestEmberContainer(fv);
-	    EmberContainerTester<double>::TestEmberContainer(dv);
+		    EmberContainerTester<double>::TestEmberContainer(dv);
-	    EmberContainerTester<float>::TestEmberContainer(fl);
+		    EmberContainerTester<float>::TestEmberContainer(fl);
-	    EmberContainerTester<double>::TestEmberContainer(dl);
+		    EmberContainerTester<double>::TestEmberContainer(dl);
-	    CopyCont(fv, fl);
+		    CopyCont(fv, fl);
 	*/
 	//QTIsaac<ISAAC_SIZE, ISAAC_INT> rand(1, 2, 3);
 	//mt19937 meow(1729);