Merge 'master' into shader-uids-awesome.

Conflicts:
	Source/Core/VideoCommon/Src/LightingShaderGen.cpp
	Source/Core/VideoCommon/Src/PixelShaderGen.cpp
	Source/Core/VideoCommon/Src/PixelShaderGen.h
	Source/Core/VideoCommon/Src/VertexShaderGen.cpp

15 files changed, 295 insertions, 48 deletions

diff --git a/Source/Core/VideoCommon/CMakeLists.txt b/Source/Core/VideoCommon/CMakeLists.txt
index 89b6828281..1bbc69a150 100644
--- a/Source/Core/VideoCommon/CMakeLists.txt
+++ b/Source/Core/VideoCommon/CMakeLists.txt
@@ -4,6 +4,7 @@ set(SRCS	Src/BPFunctions.cpp
 			Src/CPMemory.cpp
 			Src/CommandProcessor.cpp
 			Src/Debugger.cpp
+			Src/DriverDetails.cpp
 			Src/Fifo.cpp
 			Src/FPSCounter.cpp
 			Src/FramebufferManagerBase.cpp
diff --git a/Source/Core/VideoCommon/Src/BPMemory.cpp b/Source/Core/VideoCommon/Src/BPMemory.cpp
index e917632e1f..77f2f6bab3 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.cpp
+++ b/Source/Core/VideoCommon/Src/BPMemory.cpp
@@ -89,6 +89,19 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 		}
 		break;
 
+	case BPMEM_ZCOMPARE:
+		{
+			SetRegName(BPMEM_ZCOMPARE);
+			PE_CONTROL config; config.hex = cmddata;
+			const char* pixel_formats[] = { "RGB8_Z24", "RGBA6_Z24", "RGB565_Z16", "Z24", "Y8", "U8", "V8", "YUV420" };
+			const char* zformats[] = { "linear", "compressed (near)", "compressed (mid)", "compressed (far)", "inv linear", "compressed (inv near)", "compressed (inv mid)", "compressed (inv far)" };
+			snprintf(desc, desc_size, "EFB pixel format: %s\n"
+									"Depth format: %s\n"
+									"Early depth test: %s\n",
+									pixel_formats[config.pixel_format], zformats[config.zformat], no_yes[config.early_ztest]);
+		}
+		break;
+
 	case BPMEM_EFB_BR: // 0x4A
 		{
 			// TODO: Misleading name, should be BPMEM_EFB_WH instead
@@ -162,6 +175,21 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 		// TODO: Description
 		break;
 
+	case BPMEM_TX_SETIMAGE3: // 0x94
+	case BPMEM_TX_SETIMAGE3+1:
+	case BPMEM_TX_SETIMAGE3+2:
+	case BPMEM_TX_SETIMAGE3+3:
+	case BPMEM_TX_SETIMAGE3_4: // 0xB4
+	case BPMEM_TX_SETIMAGE3_4+1:
+	case BPMEM_TX_SETIMAGE3_4+2:
+	case BPMEM_TX_SETIMAGE3_4+3:
+		{
+			SetRegName(BPMEM_TX_SETIMAGE3);
+			TexImage3 teximg; teximg.hex = cmddata;
+			snprintf(desc, desc_size, "Source address (32 byte aligned): 0x%06X", teximg.image_base << 5);
+		}
+		break;
+
 	case BPMEM_TEV_COLOR_ENV: // 0xC0
 	case BPMEM_TEV_COLOR_ENV+2:
 	case BPMEM_TEV_COLOR_ENV+4:
diff --git a/Source/Core/VideoCommon/Src/DriverDetails.cpp b/Source/Core/VideoCommon/Src/DriverDetails.cpp
new file mode 100644
index 0000000000..8d3c2d9cf7
--- /dev/null
+++ b/Source/Core/VideoCommon/Src/DriverDetails.cpp
@@ -0,0 +1,69 @@
+// Copyright 2013 Dolphin Emulator Project
+// Licensed under GPLv2
+// Refer to the license.txt file included.
+
+#include <map>
+
+#include "LogManager.h"
+#include "DriverDetails.h"
+
+namespace DriverDetails
+{
+	struct BugInfo
+	{
+		Bug m_bug; // Which bug it is
+		u32 m_devfamily; // Which device(family) has the error
+		double m_versionstart; // When it started
+		double m_versionend; // When it ended
+		bool m_hasbug; // Does it have it?
+	};
+
+	// Local members
+	Vendor	m_vendor = VENDOR_UNKNOWN;
+	u32	m_devfamily = 0;
+	double	m_version = 0.0;
+
+	// This is a list of all known bugs for each vendor
+	// We use this to check if the device and driver has a issue
+	BugInfo m_qualcommbugs[] = {
+		{BUG_NODYNUBOACCESS, 300, 14.0, -1.0},
+		{BUG_BROKENCENTROID, 300, 14.0, -1.0},
+	};
+
+	std::map<std::pair<Vendor, Bug>, BugInfo> m_bugs;
+	
+	// Private function
+	void InitBugMap()
+	{
+		switch(m_vendor)
+		{
+			case VENDOR_QUALCOMM:
+				for (int a = 0; a < (sizeof(m_qualcommbugs) / sizeof(BugInfo)); ++a)
+					m_bugs[std::make_pair(m_vendor, m_qualcommbugs[a].m_bug)] = m_qualcommbugs[a];
+			break;
+			default:
+			break;
+		}
+	}
+
+	void Init(Vendor vendor, const u32 devfamily, const double version)
+	{
+		m_vendor = vendor;
+		m_devfamily = devfamily;
+		m_version = version;
+		InitBugMap();
+
+		for (auto it = m_bugs.begin(); it != m_bugs.end(); ++it)
+			if (it->second.m_devfamily == m_devfamily)
+				if (it->second.m_versionend == -1.0 || (it->second.m_versionstart <= m_version && it->second.m_versionend > m_version))
+					it->second.m_hasbug = true;
+	}
+
+	const bool HasBug(Bug bug)
+	{
+		auto it = m_bugs.find(std::make_pair(m_vendor, bug));
+		if (it == m_bugs.end())
+			return false;	
+		return it->second.m_hasbug;
+	}
+}
diff --git a/Source/Core/VideoCommon/Src/DriverDetails.h b/Source/Core/VideoCommon/Src/DriverDetails.h
new file mode 100644
index 0000000000..524825f10b
--- /dev/null
+++ b/Source/Core/VideoCommon/Src/DriverDetails.h
@@ -0,0 +1,52 @@
+// Copyright 2013 Dolphin Emulator Project
+// Licensed under GPLv2
+// Refer to the license.txt file included.
+#pragma once
+#include "CommonTypes.h"
+
+namespace DriverDetails
+{
+	// Enum of known vendors
+	// Tegra and Nvidia are separated out due to such substantial differences
+	enum Vendor
+	{
+		VENDOR_NVIDIA = 0,
+		VENDOR_ATI,
+		VENDOR_INTEL,
+		VENDOR_ARM,
+		VENDOR_QUALCOMM,
+		VENDOR_IMGTEC,
+		VENDOR_TEGRA,
+		VENDOR_VIVANTE,
+		VENDOR_UNKNOWN
+	};
+
+	// Enum of known bugs
+	// These can be vendor specific, but we put them all in here
+	// For putting a new bug in here, make sure to put a detailed comment above the enum
+	// This'll ensure we know exactly what the issue is.
+	enum Bug
+	{
+		// Bug: No Dynamic UBO array object access
+		// Affected Devices: Qualcomm/Adreno
+		// Started Version: 14
+		// Ended Version: -1
+		// Accessing UBO array members dynamically causes the Adreno shader compiler to crash
+		// Errors out with "Internal Error"
+		BUG_NODYNUBOACCESS = 0,
+		// Bug: Centroid is broken in shaders
+		// Affected devices: Qualcomm/Adreno
+		// Started Version: 14
+		// Ended Version: -1
+		// Centroid in/out, used in the shaders, is used for multisample buffers to get the texel correctly
+		// When MSAA is disabled, it acts like a regular in/out
+		// Tends to cause the driver to render full white or black
+		BUG_BROKENCENTROID,
+	};
+	
+	// Initializes our internal vendor, device family, and driver version	
+	void Init(Vendor vendor, const u32 devfamily, const double version);
+	
+	// Once Vendor and driver version is set, this will return if it has the applicable bug passed to it.
+	const bool HasBug(Bug bug);
+}
diff --git a/Source/Core/VideoCommon/Src/IndexGenerator.cpp b/Source/Core/VideoCommon/Src/IndexGenerator.cpp
index 15b2311f89..3970d8c82f 100644
--- a/Source/Core/VideoCommon/Src/IndexGenerator.cpp
+++ b/Source/Core/VideoCommon/Src/IndexGenerator.cpp
@@ -142,7 +142,7 @@ template <bool pr> void IndexGenerator::AddFan(u32 numVerts)
 	
 	if(pr)
 	{
-		for(; i<=numVerts-3; i+=3)
+		for(; i+3<=numVerts; i+=3)
 		{
 			*Tptr++ = index + i - 1;
 			*Tptr++ = index + i + 0;
@@ -153,7 +153,7 @@ template <bool pr> void IndexGenerator::AddFan(u32 numVerts)
 			numT += 3;
 		}
 		
-		for(; i<=numVerts-2; i+=2)
+		for(; i+2<=numVerts; i+=2)
 		{
 			*Tptr++ = index + i - 1;
 			*Tptr++ = index + i + 0;
diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 2aa2468eef..9b8f3a9426 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -6,6 +6,9 @@
 #include <cmath>
 #include <assert.h>
 #include <locale.h>
+#ifdef __APPLE__
+	#include <xlocale.h>
+#endif
 
 #include "LightingShaderGen.h"
 #include "PixelShaderGen.h"
@@ -213,6 +216,17 @@ static char swapModeTable[4][5];
 
 static char text[16384];
 
+// We can't use function defines since the Qualcomm shader compiler doesn't support it
+static const char *GLSLConvertFunctions[] =
+{
+	"frac", // HLSL
+	"fract", // GLSL
+	"lerp",
+	"mix"
+};
+#define FUNC_FRAC 0
+#define FUNC_LERP 2
+
 static void BuildSwapModeTable()
 {
 	static const char *swapColors = "rgba";
@@ -229,7 +243,7 @@ static void BuildSwapModeTable()
 template<class T> static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE ApiType, RegisterState RegisterStates[4]);
 template<class T> static void SampleTexture(T& out, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType);
 template<class T> static void WriteAlphaTest(T& out, pixel_shader_uid_data& uid_data, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode, bool per_pixel_depth);
-template<class T> static void WriteFog(T& out, pixel_shader_uid_data& uid_data);
+template<class T> static void WriteFog(T& out, pixel_shader_uid_data& uid_data, API_TYPE ApiType);
 
 template<class T>
 static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u32 components)
@@ -240,15 +254,23 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 										? out.template GetUidData<pixel_shader_uid_data>() : dummy_data;
 
 	out.SetBuffer(text);
+#ifndef ANDROID
+	locale_t locale;
+	locale_t old_locale;
 	if (out.GetBuffer() != NULL)
-		setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
+	{
+		locale = newlocale(LC_NUMERIC_MASK, "C", NULL); // New locale for compilation
+		old_locale = uselocale(locale); // Apply the locale for this thread
+	}
+#endif
 
 	text[sizeof(text) - 1] = 0x7C;  // canary
 
 	unsigned int numStages = bpmem.genMode.numtevstages + 1;
 	unsigned int numTexgen = bpmem.genMode.numtexgens;
 
-	bool per_pixel_depth = bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.early_ztest && bpmem.zmode.testenable;
+	bool per_pixel_depth = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.early_ztest && bpmem.zmode.testenable) || !g_ActiveConfig.bFastDepthCalc;
+	const bool bOpenGL = ApiType == API_OPENGL;
 
 	out.Write("//Pixel Shader for TEV stages\n");
 	out.Write("//%i TEV stages, %i texgens, %i IND stages\n",
@@ -267,7 +289,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 		out.Write("{\n");
 		out.Write("\tfloat z = fract( abs( x / y) ) * abs( y );\n");
 		out.Write("\treturn (x < 0) ? -z : z;\n");
-		out.Write("}\n\n");
+		out.Write("}\n");
 
 		// Declare samplers
 		for (int i = 0; i < 8; ++i)
@@ -544,7 +566,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 	}
 	// emulation of unsigned 8 overflow when casting if needed
 	if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl)
-		out.Write("\tprev = frac(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+		out.Write("\tprev = %s(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 
 	AlphaTest::TEST_RESULT Pretest = bpmem.alpha_test.TestResult();
 	uid_data.Pretest = Pretest;
@@ -552,23 +574,29 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 		WriteAlphaTest<T>(out, uid_data, ApiType, dstAlphaMode, per_pixel_depth);
 
 
-	// the screen space depth value = far z + (clip z / clip w) * z range
-	if(ApiType == API_OPENGL || ApiType == API_D3D11)
+	// D3D9 doesn't support readback of depth in pixel shader, so we always have to calculate it again.
+	// This shouldn't be a performance issue as the written depth is usually still from perspective division
+	// but this isn't true for z-textures, so there will be depth issues between enabled and disabled z-textures fragments
+	if ((ApiType == API_OPENGL || ApiType == API_D3D11) && g_ActiveConfig.bFastDepthCalc)
 		out.Write("float zCoord = rawpos.z;\n");
 	else
 	{
 		out.SetConstantsUsed(C_ZBIAS+1, C_ZBIAS+1);
-		// dx9 doesn't support 4 component position, so we have to calculate it again
+		// the screen space depth value = far z + (clip z / clip w) * z range
 		out.Write("float zCoord = " I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * " I_ZBIAS"[1].y;\n");
 	}
 
-	// Note: depth textures are disabled if early depth test is enabled
+	// depth texture can safely be ignored if the result won't be written to the depth buffer (early_ztest) and isn't used for fog either
+	bool skip_ztexture = !per_pixel_depth && !bpmem.fog.c_proj_fsel.fsel;
+
 	uid_data.ztex_op = bpmem.ztex2.op;
 	uid_data.per_pixel_depth = per_pixel_depth;
 	uid_data.fog_fsel = bpmem.fog.c_proj_fsel.fsel;
 
-	// depth texture can safely be ignored if the result won't be written to the depth buffer (early_ztest) and isn't used for fog either
-	bool skip_ztexture = !per_pixel_depth && !bpmem.fog.c_proj_fsel.fsel;
+	// Note: z-textures are not written to depth buffer if early depth test is used
+	if (per_pixel_depth && bpmem.zcontrol.early_ztest)
+		out.Write("depth = zCoord;\n");
+
 	if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !skip_ztexture)
 	{
 		// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
@@ -578,7 +606,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 
 		// U24 overflow emulation
 		out.Write("zCoord = zCoord * (16777215.0f/16777216.0f);\n");
-		out.Write("zCoord = frac(zCoord);\n");
+		out.Write("zCoord = %s(zCoord);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 		out.Write("zCoord = zCoord * (16777216.0f/16777215.0f);\n");
 
 		// Note: depth texture output is only written to depth buffer if late depth test is used
@@ -586,6 +614,8 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 		if (per_pixel_depth)
 			out.Write("depth = zCoord;\n");
 	}
+	else if (per_pixel_depth && !bpmem.zcontrol.early_ztest)
+		out.Write("depth = zCoord;\n");
 
 	if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
 	{
@@ -594,7 +624,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 	}
 	else
 	{
-		WriteFog<T>(out, uid_data);
+		WriteFog<T>(out, uid_data, ApiType);
 		out.Write("\tocol0 = prev;\n");
 	}
 
@@ -622,8 +652,13 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api
 	if (text[sizeof(text) - 1] != 0x7C)
 		PanicAlert("PixelShader generator - buffer too small, canary has been eaten!");
 
+#ifndef ANDROID
 	if (out.GetBuffer() != NULL)
-		setlocale(LC_NUMERIC, ""); // restore locale
+	{
+		uselocale(old_locale); // restore locale
+		freelocale(locale);
+	}
+#endif
 }
 
 
@@ -668,17 +703,15 @@ static const char *TEVCMPAlphaOPTable[16] =
 	"   %s.a + (abs(dot(%s.rgb, comp24) - dot(%s.rgb, comp24)) < (0.5f/255.0f) ? %s.a : 0.0f)",//#define TEVCMP_BGR24_EQ 13
 	"   %s.a + ((%s.a >= (%s.a + (0.25f/255.0f))) ? %s.a : 0.0f)",//#define TEVCMP_A8_GT 14
 	"   %s.a + (abs(%s.a - %s.a) < (0.5f/255.0f) ? %s.a : 0.0f)"//#define TEVCMP_A8_EQ 15
-
 };
 
-
 template<class T>
 static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE ApiType, RegisterState RegisterStates[4])
 {
 	int texcoord = bpmem.tevorders[n/2].getTexCoord(n&1);
 	bool bHasTexCoord = (u32)texcoord < bpmem.genMode.numtexgens;
 	bool bHasIndStage = bpmem.tevind[n].IsActive() && bpmem.tevind[n].bt < bpmem.genMode.numindstages;
-
+	bool bOpenGL = ApiType == API_OPENGL;
 	// HACK to handle cases where the tex gen is not enabled
 	if (!bHasTexCoord)
 		texcoord = 0;
@@ -813,7 +846,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 
 		char *rasswap = swapModeTable[bpmem.combiners[n].alphaC.rswap];
 		out.Write("rastemp = %s.%s;\n", tevRasTable[bpmem.tevorders[n / 2].getColorChan(n & 1)], rasswap);
-		out.Write("crastemp = frac(rastemp * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+		out.Write("crastemp = %s(rastemp * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 	}
 
 
@@ -855,7 +888,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 		out.Write("konsttemp = float4(%s, %s);\n", tevKSelTableC[kc], tevKSelTableA[ka]);
 		if(kc > 7 || ka > 7)
 		{
-			out.Write("ckonsttemp = frac(konsttemp * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			out.Write("ckonsttemp = %s(konsttemp * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 		}
 		else
 		{
@@ -874,7 +907,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 	{
 		if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl)
 		{
-			out.Write("cprev = frac(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			out.Write("cprev = %s(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 			RegisterStates[0].AlphaNeedOverflowControl = false;
 			RegisterStates[0].ColorNeedOverflowControl = false;
 		}
@@ -893,7 +926,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 		out.SetConstantsUsed(C_COLORS+1,C_COLORS+1);
 		if(RegisterStates[1].AlphaNeedOverflowControl || RegisterStates[1].ColorNeedOverflowControl)
 		{
-			out.Write("cc0 = frac(c0 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			out.Write("cc0 = %s(c0 * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 			RegisterStates[1].AlphaNeedOverflowControl = false;
 			RegisterStates[1].ColorNeedOverflowControl = false;
 		}
@@ -912,7 +945,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 		out.SetConstantsUsed(C_COLORS+2,C_COLORS+2);
 		if(RegisterStates[2].AlphaNeedOverflowControl || RegisterStates[2].ColorNeedOverflowControl)
 		{
-			out.Write("cc1 = frac(c1 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			out.Write("cc1 = frac(c1 * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 			RegisterStates[2].AlphaNeedOverflowControl = false;
 			RegisterStates[2].ColorNeedOverflowControl = false;
 		}
@@ -931,7 +964,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 		out.SetConstantsUsed(C_COLORS+3,C_COLORS+3);
 		if(RegisterStates[3].AlphaNeedOverflowControl || RegisterStates[3].ColorNeedOverflowControl)
 		{
-			out.Write("cc2 = frac(c2 * (255.0f/256.0f)) * (256.0f/255.0f);\n");
+			out.Write("cc2 = %s(c2 * (255.0f/256.0f)) * (256.0f/255.0f);\n", GLSLConvertFunctions[FUNC_FRAC + bOpenGL]);
 			RegisterStates[3].AlphaNeedOverflowControl = false;
 			RegisterStates[3].ColorNeedOverflowControl = false;
 		}
@@ -962,7 +995,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 
 	out.Write("// color combine\n");
 	if (cc.clamp)
-		out.Write("%s = saturate(", tevCOutputTable[cc.dest]);
+		out.Write("%s = clamp(", tevCOutputTable[cc.dest]);
 	else
 		out.Write("%s = ", tevCOutputTable[cc.dest]);
 
@@ -987,7 +1020,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 		else if (cc.b == TEVCOLORARG_ZERO)
 			out.Write("%s*(float3(1.0f, 1.0f, 1.0f)-%s)", tevCInputTable[cc.a + 16], tevCInputTable[cc.c + 16]);
 		else
-			out.Write("lerp(%s, %s, %s)", tevCInputTable[cc.a + 16], tevCInputTable[cc.b + 16], tevCInputTable[cc.c + 16]);
+			out.Write("%s(%s, %s, %s)", GLSLConvertFunctions[FUNC_LERP + bOpenGL], tevCInputTable[cc.a + 16], tevCInputTable[cc.b + 16], tevCInputTable[cc.c + 16]);
 
 		out.Write("%s", tevBiasTable[cc.bias]);
 
@@ -1004,7 +1037,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 				tevCInputTable[cc.c + 16]);
 	}
 	if (cc.clamp)
-		out.Write(")");
+		out.Write(", 0.0, 1.0)");
 	out.Write(";\n");
 
 	RegisterStates[ac.dest].AlphaNeedOverflowControl = (ac.clamp == 0);
@@ -1012,7 +1045,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 
 	out.Write("// alpha combine\n");
 	if (ac.clamp)
-		out.Write("%s = saturate(", tevAOutputTable[ac.dest]);
+		out.Write("%s = clamp(", tevAOutputTable[ac.dest]);
 	else
 		out.Write("%s = ", tevAOutputTable[ac.dest]);
 
@@ -1034,7 +1067,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 		else if (ac.b == TEVALPHAARG_ZERO)
 			out.Write("%s.a*(1.0f-%s.a)", tevAInputTable[ac.a + 8], tevAInputTable[ac.c + 8]);
 		else
-			out.Write("lerp(%s.a, %s.a, %s.a)", tevAInputTable[ac.a + 8], tevAInputTable[ac.b + 8], tevAInputTable[ac.c + 8]);
+			out.Write("%s(%s.a, %s.a, %s.a)", GLSLConvertFunctions[FUNC_LERP + bOpenGL], tevAInputTable[ac.a + 8], tevAInputTable[ac.b + 8], tevAInputTable[ac.c + 8]);
 
 		out.Write("%s",tevBiasTable[ac.bias]);
 
@@ -1053,7 +1086,7 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE
 				tevAInputTable[ac.c + 8]);
 	}
 	if (ac.clamp)
-		out.Write(")");
+		out.Write(", 0.0, 1.0)");
 	out.Write(";\n\n");
 	out.Write("// TEV done\n");
 }
@@ -1157,8 +1190,10 @@ static const char *tevFogFuncsTable[] =
 };
 
 template<class T>
-static void WriteFog(T& out, pixel_shader_uid_data& uid_data)
+static void WriteFog(T& out, pixel_shader_uid_data& uid_data, API_TYPE ApiType)
 {
+	bool bOpenGL = ApiType == API_OPENGL;
+
 	uid_data.fog_fsel = bpmem.fog.c_proj_fsel.fsel;
 	if(bpmem.fog.c_proj_fsel.fsel == 0)
 		return; // no Fog
@@ -1191,7 +1226,7 @@ static void WriteFog(T& out, pixel_shader_uid_data& uid_data)
 		out.Write("\tze *= x_adjust;\n");
 	}
 
-	out.Write("\tfloat fog = saturate(ze - " I_FOG"[1].z);\n");
+	out.Write("\tfloat fog = clamp(ze - " I_FOG"[1].z, 0.0, 1.0);\n");
 
 	if (bpmem.fog.c_proj_fsel.fsel > 3)
 	{
@@ -1203,7 +1238,7 @@ static void WriteFog(T& out, pixel_shader_uid_data& uid_data)
 			WARN_LOG(VIDEO, "Unknown Fog Type! %08x", bpmem.fog.c_proj_fsel.fsel);
 	}
 
-	out.Write("\tprev.rgb = lerp(prev.rgb, " I_FOG"[0].rgb, fog);\n");
+	out.Write("\tprev.rgb = %s(prev.rgb, " I_FOG"[0].rgb, fog);\n", GLSLConvertFunctions[FUNC_LERP + bOpenGL]);
 }
 
 void GetPixelShaderUid(PixelShaderUid& object, DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType, u32 components)
diff --git a/Source/Core/VideoCommon/Src/Statistics.cpp b/Source/Core/VideoCommon/Src/Statistics.cpp
index 3dcd281b7f..0c69499355 100644
--- a/Source/Core/VideoCommon/Src/Statistics.cpp
+++ b/Source/Core/VideoCommon/Src/Statistics.cpp
@@ -55,6 +55,9 @@ char *Statistics::ToString(char *ptr)
 	ptr+=sprintf(ptr,"CP loads (DL): %i\n",stats.thisFrame.numCPLoadsInDL);
 	ptr+=sprintf(ptr,"BP loads: %i\n",stats.thisFrame.numBPLoads);
 	ptr+=sprintf(ptr,"BP loads (DL): %i\n",stats.thisFrame.numBPLoadsInDL);
+	ptr+=sprintf(ptr,"Vertex streamed: %i kB\n",stats.thisFrame.bytesVertexStreamed/1024);
+	ptr+=sprintf(ptr,"Index streamed: %i kB\n",stats.thisFrame.bytesIndexStreamed/1024);
+	ptr+=sprintf(ptr,"Uniform streamed: %i kB\n",stats.thisFrame.bytesUniformStreamed/1024);
 	ptr+=sprintf(ptr,"Vertex Loaders: %i\n",stats.numVertexLoaders);
 
 	std::string text1;
diff --git a/Source/Core/VideoCommon/Src/Statistics.h b/Source/Core/VideoCommon/Src/Statistics.h
index cb8a3afdc5..a7976651a5 100644
--- a/Source/Core/VideoCommon/Src/Statistics.h
+++ b/Source/Core/VideoCommon/Src/Statistics.h
@@ -60,6 +60,10 @@ struct Statistics
 		int numBufferSplits;
 
 		int numDListsCalled;
+		
+		int bytesVertexStreamed;
+		int bytesIndexStreamed;
+		int bytesUniformStreamed;
 	};
 	ThisFrame thisFrame;
 	void ResetFrame();
diff --git a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
index d5d21f89d0..7d5b9354ad 100644
--- a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
+++ b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
@@ -32,6 +32,7 @@ TextureCache::TexCache TextureCache::textures;
 
 TextureCache::BackupConfig TextureCache::backup_config;
 
+bool invalidate_texture_cache_requested;
 
 TextureCache::TCacheEntryBase::~TCacheEntryBase()
 {
@@ -49,6 +50,13 @@ TextureCache::TextureCache()
 		HiresTextures::Init(SConfig::GetInstance().m_LocalCoreStartupParameter.m_strUniqueID.c_str());
 
 	SetHash64Function(g_ActiveConfig.bHiresTextures || g_ActiveConfig.bDumpTextures);
+
+	invalidate_texture_cache_requested = false;
+}
+
+void TextureCache::RequestInvalidateTextureCache()
+{
+	invalidate_texture_cache_requested = true;
 }
 
 void TextureCache::Invalidate()
@@ -80,7 +88,8 @@ void TextureCache::OnConfigChanged(VideoConfig& config)
 		if (config.iSafeTextureCache_ColorSamples != backup_config.s_colorsamples ||
 			config.bTexFmtOverlayEnable != backup_config.s_texfmt_overlay ||
 			config.bTexFmtOverlayCenter != backup_config.s_texfmt_overlay_center ||
-			config.bHiresTextures != backup_config.s_hires_textures)
+			config.bHiresTextures != backup_config.s_hires_textures ||
+			invalidate_texture_cache_requested)
 		{
 			g_texture_cache->Invalidate();
 
@@ -89,6 +98,8 @@ void TextureCache::OnConfigChanged(VideoConfig& config)
 
 			SetHash64Function(g_ActiveConfig.bHiresTextures || g_ActiveConfig.bDumpTextures);
 			TexDecoder_SetTexFmtOverlayOptions(g_ActiveConfig.bTexFmtOverlayEnable, g_ActiveConfig.bTexFmtOverlayCenter);
+
+			invalidate_texture_cache_requested = false;
 		}
 
 		// TODO: Probably shouldn't clear all render targets here, just mark them dirty or something.
diff --git a/Source/Core/VideoCommon/Src/TextureCacheBase.h b/Source/Core/VideoCommon/Src/TextureCacheBase.h
index acfb2e94b2..1afcb8cd2d 100644
--- a/Source/Core/VideoCommon/Src/TextureCacheBase.h
+++ b/Source/Core/VideoCommon/Src/TextureCacheBase.h
@@ -107,6 +107,8 @@ public:
 	static void CopyRenderTargetToTexture(u32 dstAddr, unsigned int dstFormat, unsigned int srcFormat,
 		const EFBRectangle& srcRect, bool isIntensity, bool scaleByHalf);
 
+	static void RequestInvalidateTextureCache();
+
 protected:
 	TextureCache();
 
@@ -118,7 +120,6 @@ private:
 	static PC_TexFormat LoadCustomTexture(u64 tex_hash, int texformat, unsigned int level, unsigned int& width, unsigned int& height);
 	static void DumpTexture(TCacheEntryBase* entry, unsigned int level);
 
-
 	typedef std::map<u32, TCacheEntryBase*> TexCache;
 
 	static TexCache textures;
diff --git a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
index 539d8799f9..40873b903e 100644
--- a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
+++ b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp
@@ -6,6 +6,9 @@
 #include <stdio.h>
 #include <math.h>
 #include <locale.h>
+#ifdef __APPLE__
+	#include <xlocale.h>
+#endif
 
 #include "TextureConversionShader.h"
 #include "TextureDecoder.h"
@@ -804,7 +807,10 @@ void WriteZ24Encoder(char* p, API_TYPE ApiType)
 
 const char *GenerateEncodingShader(u32 format,API_TYPE ApiType)
 {
-	setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
+#ifndef ANDROID
+	locale_t locale = newlocale(LC_NUMERIC_MASK, "C", NULL); // New locale for compilation
+	locale_t old_locale = uselocale(locale); // Apply the locale for this thread
+#endif
 	text[sizeof(text) - 1] = 0x7C;  // canary
 
 	char *p = text;
@@ -887,8 +893,11 @@ const char *GenerateEncodingShader(u32 format,API_TYPE ApiType)
 
 	if (text[sizeof(text) - 1] != 0x7C)
 		PanicAlert("TextureConversionShader generator - buffer too small, canary has been eaten!");
-
-	setlocale(LC_NUMERIC, ""); // restore locale
+	
+#ifndef ANDROID
+	uselocale(old_locale); // restore locale
+	freelocale(locale);
+#endif
 	return text;
 }
 
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
index 6cbf273097..3edb51c505 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
@@ -4,11 +4,15 @@
 
 #include <math.h>
 #include <locale.h>
+#ifdef __APPLE__
+	#include <xlocale.h>
+#endif
 
 #include "NativeVertexFormat.h"
 
 #include "BPMemory.h"
 #include "CPMemory.h"
+#include "DriverDetails.h"
 #include "LightingShaderGen.h"
 #include "VertexShaderGen.h"
 #include "VideoConfig.h"
@@ -71,9 +75,15 @@ static void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
 											? out.template GetUidData<vertex_shader_uid_data>() : dummy_data;
 
 	out.SetBuffer(text);
+#ifndef ANDROID
+	locale_t locale;
+	locale_t old_locale;
 	if (out.GetBuffer() != NULL)
-		setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
-
+	{
+		locale = newlocale(LC_NUMERIC_MASK, "C", NULL); // New locale for compilation
+		old_locale = uselocale(locale); // Apply the locale for this thread
+	}
+#endif
 	text[sizeof(text) - 1] = 0x7C;  // canary
 
 	_assert_(bpmem.genMode.numtexgens == xfregs.numTexGen.numTexGens);
@@ -214,13 +224,22 @@ static void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
 			out.Write("int posmtx = int(fposmtx);\n");
 		}
 
-		out.Write("float4 pos = float4(dot(" I_TRANSFORMMATRICES"[posmtx], rawpos), dot(" I_TRANSFORMMATRICES"[posmtx+1], rawpos), dot(" I_TRANSFORMMATRICES"[posmtx+2], rawpos), 1);\n");		
-
-		if (components & VB_HAS_NRMALL) {
-			out.Write("int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n");
-			out.Write("float3 N0 = " I_NORMALMATRICES"[normidx].xyz, N1 = " I_NORMALMATRICES"[normidx+1].xyz, N2 = " I_NORMALMATRICES"[normidx+2].xyz;\n");
+		if (DriverDetails::HasBug(DriverDetails::BUG_NODYNUBOACCESS))
+		{
+			// This'll cause issues, but  it can't be helped
+			out.Write("float4 pos = float4(dot(" I_TRANSFORMMATRICES"[0], rawpos), dot(" I_TRANSFORMMATRICES"[1], rawpos), dot(" I_TRANSFORMMATRICES"[2], rawpos), 1);\n");
+			if (components & VB_HAS_NRMALL)
+				out.Write("float3 N0 = " I_NORMALMATRICES"[0].xyz, N1 = " I_NORMALMATRICES"[1].xyz, N2 = " I_NORMALMATRICES"[2].xyz;\n");
 		}
+		else
+		{
+			out.Write("float4 pos = float4(dot(" I_TRANSFORMMATRICES"[posmtx], rawpos), dot(" I_TRANSFORMMATRICES"[posmtx+1], rawpos), dot(" I_TRANSFORMMATRICES"[posmtx+2], rawpos), 1);\n");
 
+			if (components & VB_HAS_NRMALL) {
+				out.Write("int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n");
+				out.Write("float3 N0 = " I_NORMALMATRICES"[normidx].xyz, N1 = " I_NORMALMATRICES"[normidx+1].xyz, N2 = " I_NORMALMATRICES"[normidx+2].xyz;\n");
+			}
+		}
 		if (components & VB_HAS_NRM0)
 			out.Write("float3 _norm0 = normalize(float3(dot(N0, rawnorm0), dot(N1, rawnorm0), dot(N2, rawnorm0)));\n");
 		if (components & VB_HAS_NRM1)
@@ -531,8 +550,13 @@ static void GenerateVertexShader(T& out, u32 components, API_TYPE api_type)
 	if (text[sizeof(text) - 1] != 0x7C)
 		PanicAlert("VertexShader generator - buffer too small, canary has been eaten!");
 
+#ifndef ANDROID
 	if (out.GetBuffer() != NULL)
-		setlocale(LC_NUMERIC, ""); // restore locale
+	{
+		uselocale(old_locale); // restore locale
+		freelocale(locale);
+	}
+#endif
 }
 
 void GetVertexShaderUid(VertexShaderUid& object, u32 components, API_TYPE api_type)
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.cpp b/Source/Core/VideoCommon/Src/VideoConfig.cpp
index 0b70b5440f..7827c0008a 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.cpp
+++ b/Source/Core/VideoCommon/Src/VideoConfig.cpp
@@ -10,12 +10,15 @@
 #include "VideoCommon.h"
 #include "FileUtil.h"
 #include "Core.h"
+#include "Movie.h"
 
 VideoConfig g_Config;
 VideoConfig g_ActiveConfig;
 
 void UpdateActiveConfig()
 {
+	if (Movie::IsPlayingInput() && Movie::IsConfigSaved())
+		Movie::SetGraphicsConfig();
 	g_ActiveConfig = g_Config;
 }
 
@@ -64,6 +67,7 @@ void VideoConfig::Load(const char *ini_file)
 	iniFile.Get("Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle, 0);
 	iniFile.Get("Settings", "EnablePixelLighting", &bEnablePixelLighting, 0);
 	iniFile.Get("Settings", "HackedBufferUpload", &bHackedBufferUpload, 0);
+	iniFile.Get("Settings", "FastDepthCalc", &bFastDepthCalc, true);
 
 	iniFile.Get("Settings", "MSAA", &iMultisampleMode, 0);
 	iniFile.Get("Settings", "EFBScale", &iEFBScale, (int) SCALE_1X); // native
@@ -123,6 +127,7 @@ void VideoConfig::GameIniLoad(const char *ini_file)
 	iniFile.GetIfExists("Video_Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle);
 	iniFile.GetIfExists("Video_Settings", "EnablePixelLighting", &bEnablePixelLighting);
 	iniFile.GetIfExists("Video_Settings", "HackedBufferUpload", &bHackedBufferUpload);
+	iniFile.GetIfExists("Video_Settings", "FastDepthCalc", &bFastDepthCalc);
 	iniFile.GetIfExists("Video_Settings", "MSAA", &iMultisampleMode);
 	int tmp = -9000;
 	iniFile.GetIfExists("Video_Settings", "EFBScale", &tmp); // integral
@@ -219,6 +224,7 @@ void VideoConfig::Save(const char *ini_file)
 	iniFile.Set("Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	iniFile.Set("Settings", "EnablePixelLighting", bEnablePixelLighting);
 	iniFile.Set("Settings", "HackedBufferUpload", bHackedBufferUpload);
+	iniFile.Set("Settings", "FastDepthCalc", bFastDepthCalc);
 
 	iniFile.Set("Settings", "ShowEFBCopyRegions", bShowEFBCopyRegions);
 	iniFile.Set("Settings", "MSAA", iMultisampleMode);
@@ -283,6 +289,7 @@ void VideoConfig::GameIniSave(const char* default_ini, const char* game_ini)
 	SET_IF_DIFFERS("Video_Settings", "AnaglyphStereoSeparation", iAnaglyphStereoSeparation);
 	SET_IF_DIFFERS("Video_Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	SET_IF_DIFFERS("Video_Settings", "EnablePixelLighting", bEnablePixelLighting);
+	SET_IF_DIFFERS("Video_Settings", "FastDepthCalc", bFastDepthCalc);
 	SET_IF_DIFFERS("Video_Settings", "MSAA", iMultisampleMode);
 	SET_IF_DIFFERS("Video_Settings", "EFBScale", iEFBScale); // integral
 	SET_IF_DIFFERS("Video_Settings", "DstAlphaPass", bDstAlphaPass);
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.h b/Source/Core/VideoCommon/Src/VideoConfig.h
index 9b5076df8c..ec4ad3ed3d 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.h
+++ b/Source/Core/VideoCommon/Src/VideoConfig.h
@@ -123,6 +123,7 @@ struct VideoConfig
 	bool bUseBBox;
 	bool bEnablePixelLighting;
 	bool bHackedBufferUpload;
+	bool bFastDepthCalc;
 
 	int iLog; // CONF_ bits
 	int iSaveTargetId; // TODO: Should be dropped
diff --git a/Source/Core/VideoCommon/VideoCommon.vcxproj b/Source/Core/VideoCommon/VideoCommon.vcxproj
index 1baf417797..617e4ec567 100644
--- a/Source/Core/VideoCommon/VideoCommon.vcxproj
+++ b/Source/Core/VideoCommon/VideoCommon.vcxproj
@@ -182,6 +182,7 @@
     <ClCompile Include="Src\CommandProcessor.cpp" />
     <ClCompile Include="Src\CPMemory.cpp" />
     <ClCompile Include="Src\Debugger.cpp" />
+    <ClCompile Include="Src\DriverDetails.cpp" />
     <ClCompile Include="Src\EmuWindow.cpp" />
     <ClCompile Include="Src\Fifo.cpp" />
     <ClCompile Include="Src\FPSCounter.cpp" />
@@ -229,6 +230,7 @@
     <ClInclude Include="Src\CPMemory.h" />
     <ClInclude Include="Src\DataReader.h" />
     <ClInclude Include="Src\Debugger.h" />
+    <ClInclude Include="Src\DriverDetails.h" />
     <ClInclude Include="Src\DLCache.h" />
     <ClInclude Include="Src\EmuWindow.h" />
     <ClInclude Include="Src\Fifo.h" />
@@ -303,4 +305,4 @@
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
   <ImportGroup Label="ExtensionTargets">
   </ImportGroup>
-</Project>
\ No newline at end of file
+</Project>