Merge branch 'immediate-removal' into GLSL-master

Conflicts:
	Source/Core/VideoCommon/Src/PixelShaderGen.cpp
	Source/Plugins/Plugin_VideoSoftware/Src/SWRenderer.cpp

immediate-removal is a new created branch seperated from master but reverted the revert of immediate-removal
so we get less conflicts by merging

19 files changed, 519 insertions, 345 deletions

diff --git a/Source/Core/VideoCommon/Src/BPMemory.cpp b/Source/Core/VideoCommon/Src/BPMemory.cpp
index c452b238b6..38489bcf7a 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.cpp
+++ b/Source/Core/VideoCommon/Src/BPMemory.cpp
@@ -55,6 +55,11 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 		snprintf(name, name_size, #reg); \
 		(void)(reg);
 
+	case BPMEM_GENMODE: // 0x00
+		SetRegName(BPMEM_GENMODE);
+		// TODO: Description
+		break;
+
 	case BPMEM_DISPLAYCOPYFILER: // 0x01
 		// TODO: This is actually the sample pattern used for copies from an antialiased EFB
 		SetRegName(BPMEM_DISPLAYCOPYFILER);
@@ -75,6 +80,28 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 		}
 		break;
 
+	case BPMEM_BLENDMODE: // 0x41
+		{
+			SetRegName(BPMEM_BLENDMODE);
+			BlendMode mode; mode.hex = cmddata;
+			const char* dstfactors[] = { "0", "1", "src_color", "1-src_color", "src_alpha", "1-src_alpha", "dst_alpha", "1-dst_alpha" };
+			const char* srcfactors[] = { "0", "1", "dst_color", "1-dst_color", "src_alpha", "1-src_alpha", "dst_alpha", "1-dst_alpha" };
+			const char* logicmodes[] = { "0", "s & d", "s & ~d", "s", "~s & d", "d", "s ^ d", "s | d", "~(s | d)", "~(s ^ d)", "~d", "s | ~d", "~s", "~s | d", "~(s & d)", "1" };
+			snprintf(desc, desc_size, "Enable: %s\n"
+									"Logic ops: %s\n"
+									"Dither: %s\n"
+									"Color write: %s\n"
+									"Alpha write: %s\n"
+									"Dest factor: %s\n"
+									"Source factor: %s\n"
+									"Subtract: %s\n"
+									"Logic mode: %s\n",
+									no_yes[mode.blendenable], no_yes[mode.logicopenable], no_yes[mode.dither],
+									no_yes[mode.colorupdate], no_yes[mode.alphaupdate], dstfactors[mode.dstfactor],
+									srcfactors[mode.srcfactor], no_yes[mode.subtract], logicmodes[mode.logicmode]);
+		}
+		break;
+
 	case BPMEM_EFB_BR: // 0x4A
 		{
 			// TODO: Misleading name, should be BPMEM_EFB_WH instead
@@ -148,6 +175,147 @@ void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size
 		// TODO: Description
 		break;
 
-#undef SET_REG_NAME
+	case BPMEM_TEV_COLOR_ENV: // 0xC0
+	case BPMEM_TEV_COLOR_ENV+2:
+	case BPMEM_TEV_COLOR_ENV+4:
+	case BPMEM_TEV_COLOR_ENV+8:
+	case BPMEM_TEV_COLOR_ENV+10:
+	case BPMEM_TEV_COLOR_ENV+12:
+	case BPMEM_TEV_COLOR_ENV+14:
+	case BPMEM_TEV_COLOR_ENV+16:
+	case BPMEM_TEV_COLOR_ENV+18:
+	case BPMEM_TEV_COLOR_ENV+20:
+	case BPMEM_TEV_COLOR_ENV+22:
+	case BPMEM_TEV_COLOR_ENV+24:
+	case BPMEM_TEV_COLOR_ENV+26:
+	case BPMEM_TEV_COLOR_ENV+28:
+	case BPMEM_TEV_COLOR_ENV+30:
+		{
+			SetRegName(BPMEM_TEV_COLOR_ENV);
+			TevStageCombiner::ColorCombiner cc; cc.hex = cmddata;
+			const char* tevin[] =
+			{
+				"prev.rgb", "prev.aaa",
+				"c0.rgb", "c0.aaa",
+				"c1.rgb", "c1.aaa",
+				"c2.rgb", "c2.aaa",
+				"tex.rgb", "tex.aaa",
+				"ras.rgb", "ras.aaa",
+				"ONE", "HALF", "konst.rgb", "ZERO",
+			};
+			const char* tevbias[] = { "0", "+0.5", "-0.5", "compare" };
+			const char* tevop[] = { "add", "sub" };
+			const char* tevscale[] = { "1", "2", "4", "0.5" };
+			const char* tevout[] = { "prev.rgb", "c0.rgb", "c1.rgb", "c2.rgb" };
+			snprintf(desc, desc_size, "tev stage: %d\n"
+									"a: %s\n"
+									"b: %s\n"
+									"c: %s\n"
+									"d: %s\n"
+									"bias: %s\n"
+									"op: %s\n"
+									"clamp: %s\n"
+									"scale factor: %s\n"
+									"dest: %s\n",
+									(data[0] - BPMEM_TEV_COLOR_ENV)/2, tevin[cc.a], tevin[cc.b], tevin[cc.c], tevin[cc.d],
+									tevbias[cc.bias], tevop[cc.op], no_yes[cc.clamp], tevscale[cc.shift], tevout[cc.dest]);
+			break;
+		}
+
+	case BPMEM_TEV_ALPHA_ENV: // 0xC1
+	case BPMEM_TEV_ALPHA_ENV+2:
+	case BPMEM_TEV_ALPHA_ENV+4:
+	case BPMEM_TEV_ALPHA_ENV+6:
+	case BPMEM_TEV_ALPHA_ENV+8:
+	case BPMEM_TEV_ALPHA_ENV+10:
+	case BPMEM_TEV_ALPHA_ENV+12:
+	case BPMEM_TEV_ALPHA_ENV+14:
+	case BPMEM_TEV_ALPHA_ENV+16:
+	case BPMEM_TEV_ALPHA_ENV+18:
+	case BPMEM_TEV_ALPHA_ENV+20:
+	case BPMEM_TEV_ALPHA_ENV+22:
+	case BPMEM_TEV_ALPHA_ENV+24:
+	case BPMEM_TEV_ALPHA_ENV+26:
+	case BPMEM_TEV_ALPHA_ENV+28:
+	case BPMEM_TEV_ALPHA_ENV+30:
+		{
+			SetRegName(BPMEM_TEV_ALPHA_ENV);
+			TevStageCombiner::AlphaCombiner ac; ac.hex = cmddata;
+			const char* tevin[] =
+			{
+				"prev", "c0", "c1", "c2",
+				"tex", "ras", "konst", "ZERO",
+			};
+			const char* tevbias[] = { "0", "+0.5", "-0.5", "compare" };
+			const char* tevop[] = { "add", "sub" };
+			const char* tevscale[] = { "1", "2", "4", "0.5" };
+			const char* tevout[] = { "prev", "c0", "c1", "c2" };
+			snprintf(desc, desc_size, "tev stage: %d\n"
+									"a: %s\n"
+									"b: %s\n"
+									"c: %s\n"
+									"d: %s\n"
+									"bias: %s\n"
+									"op: %s\n"
+									"clamp: %s\n"
+									"scale factor: %s\n"
+									"dest: %s\n"
+									"ras sel: %d\n"
+									"tex sel: %d\n",
+									(data[0] - BPMEM_TEV_ALPHA_ENV)/2, tevin[ac.a], tevin[ac.b], tevin[ac.c], tevin[ac.d],
+									tevbias[ac.bias], tevop[ac.op], no_yes[ac.clamp], tevscale[ac.shift], tevout[ac.dest],
+									ac.rswap, ac.tswap);
+			break;
+		}
+
+		case BPMEM_ALPHACOMPARE: // 0xF3
+			{
+				SetRegName(BPMEM_ALPHACOMPARE);
+				AlphaTest test; test.hex = cmddata;
+				const char* functions[] = { "NEVER", "LESS", "EQUAL", "LEQUAL", "GREATER", "NEQUAL", "GEQUAL", "ALWAYS" };
+				const char* logic[] = { "AND", "OR", "XOR", "XNOR" };
+				snprintf(desc, desc_size, "test 1: %s (ref: %#02x)\n"
+										"test 2: %s (ref: %#02x)\n"
+										"logic: %s\n",
+										functions[test.comp0], test.ref0, functions[test.comp1], test.ref1, logic[test.logic]);
+				break;
+			}
+
+#undef SetRegName
+	}
+}
+
+AlphaTest::TEST_RESULT AlphaTest::TestResult()
+{
+	switch(logic)
+	{
+	case 0: // AND
+		if (comp0 == ALPHACMP_ALWAYS && comp1 == ALPHACMP_ALWAYS)
+			return PASS;
+		if (comp0 == ALPHACMP_NEVER || comp1 == ALPHACMP_NEVER)
+			return FAIL;
+		break;
+
+	case 1: // OR
+		if (comp0 == ALPHACMP_ALWAYS || comp1 == ALPHACMP_ALWAYS)
+			return PASS;
+		if (comp0 == ALPHACMP_NEVER && comp1 == ALPHACMP_NEVER)
+			return FAIL;
+		break;
+
+	case 2: // XOR
+		if ((comp0 == ALPHACMP_ALWAYS && comp1 == ALPHACMP_NEVER) || (comp0 == ALPHACMP_NEVER && comp1 == ALPHACMP_ALWAYS))
+			return PASS;
+		if ((comp0 == ALPHACMP_ALWAYS && comp1 == ALPHACMP_ALWAYS) || (comp0 == ALPHACMP_NEVER && comp1 == ALPHACMP_NEVER))
+			return FAIL;
+		break;
+
+	case 3: // XNOR
+		if ((comp0 == ALPHACMP_ALWAYS && comp1 == ALPHACMP_NEVER) || (comp0 == ALPHACMP_NEVER && comp1 == ALPHACMP_ALWAYS))
+			return FAIL;
+		if ((comp0 == ALPHACMP_ALWAYS && comp1 == ALPHACMP_ALWAYS) || (comp0 == ALPHACMP_NEVER && comp1 == ALPHACMP_NEVER))
+			return PASS;
+		break;
 	}
+	return UNDETERMINED;
 }
diff --git a/Source/Core/VideoCommon/Src/BPMemory.h b/Source/Core/VideoCommon/Src/BPMemory.h
index 9284674aa1..85e95f4f8b 100644
--- a/Source/Core/VideoCommon/Src/BPMemory.h
+++ b/Source/Core/VideoCommon/Src/BPMemory.h
@@ -793,7 +793,7 @@ union PE_CONTROL
     {
         u32 pixel_format : 3;	// PIXELFMT_X
         u32 zformat : 3;		// Z Compression for 16bit Z format
-        u32 zcomploc : 1;		// 1: before tex stage
+        u32 early_ztest : 1;	// 1: before tex stage
         u32 unused : 17;
         u32 rid : 8;
     };
@@ -857,7 +857,7 @@ union TevKSel
     int getKA(int i) {return i?kasel1:kasel0;}
 };
 
-union AlphaFunc
+union AlphaTest
 {
     struct
     {
@@ -868,6 +868,15 @@ union AlphaFunc
         u32 logic : 2;
     };
     u32 hex;
+
+	enum TEST_RESULT
+	{
+		UNDETERMINED = 0,
+		FAIL = 1,
+		PASS = 2,
+	};
+
+	TEST_RESULT TestResult();
 };
 
 union UPE_Copy
@@ -981,7 +990,7 @@ struct BPMemory
     TevReg tevregs[4];  //0xE0
     FogRangeParams fogRange;
     FogParams fog; //0xEE,0xEF,0xF0,0xF1,0xF2
-    AlphaFunc alphaFunc; //0xF3
+    AlphaTest alpha_test; //0xF3
     ZTex1 ztex1; //0xf4,0xf5
     ZTex2 ztex2;
     TevKSel tevksel[8];//0xf6,0xf7,f8,f9,fa,fb,fc,fd
diff --git a/Source/Core/VideoCommon/Src/BPStructs.cpp b/Source/Core/VideoCommon/Src/BPStructs.cpp
index 47dab0ec94..3d48d946a6 100644
--- a/Source/Core/VideoCommon/Src/BPStructs.cpp
+++ b/Source/Core/VideoCommon/Src/BPStructs.cpp
@@ -359,9 +359,9 @@ void BPWritten(const BPCmd& bp)
 			PixelShaderManager::SetFogColorChanged();
 		break;
 	case BPMEM_ALPHACOMPARE: // Compare Alpha Values
-		PRIM_LOG("alphacmp: ref0=%d, ref1=%d, comp0=%d, comp1=%d, logic=%d", bpmem.alphaFunc.ref0,
-				bpmem.alphaFunc.ref1, bpmem.alphaFunc.comp0, bpmem.alphaFunc.comp1, bpmem.alphaFunc.logic);
-		PixelShaderManager::SetAlpha(bpmem.alphaFunc);
+		PRIM_LOG("alphacmp: ref0=%d, ref1=%d, comp0=%d, comp1=%d, logic=%d", bpmem.alpha_test.ref0,
+				bpmem.alpha_test.ref1, bpmem.alpha_test.comp0, bpmem.alpha_test.comp1, bpmem.alpha_test.logic);
+		PixelShaderManager::SetAlpha(bpmem.alpha_test);
 		break;
 	case BPMEM_BIAS: // BIAS
 		PRIM_LOG("ztex bias=0x%x", bpmem.ztex1.bias);
@@ -501,18 +501,40 @@ void BPWritten(const BPCmd& bp)
 		// if this is different from 0, manual TMEM management is used (GX_PreloadEntireTexture).
 		if (bp.newvalue != 0)
 		{
-			// NOTE(neobrain): Apparently tmemodd doesn't affect hardware behavior at all (libogc uses it just as a buffer and switches its contents with tmemeven whenever this is called)
+			// TODO: Not quite sure if this is completely correct (likely not)
+			// NOTE: libogc's implementation of GX_PreloadEntireTexture seems flawed, so it's not necessarily a good reference for RE'ing this feature.
+
 			BPS_TmemConfig& tmem_cfg = bpmem.tmem_config;
-			u8* ram_ptr = Memory::GetPointer(tmem_cfg.preload_addr << 5);
-			u32 tmem_addr = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE;
-			u32 size = tmem_cfg.preload_tile_info.count * 32;
+			u8* src_ptr = Memory::GetPointer(tmem_cfg.preload_addr << 5); // TODO: Should we add mask here on GC?
+			u32 size = tmem_cfg.preload_tile_info.count * TMEM_LINE_SIZE;
+			u32 tmem_addr_even = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE;
+
+			if (tmem_cfg.preload_tile_info.type != 3)
+			{
+				if (tmem_addr_even + size > TMEM_SIZE)
+					size = TMEM_SIZE - tmem_addr_even;
 
-			// Check if the game has overflowed TMEM, and copy up to the limit.
-			// Paper Mario does this when entering the Great Boogly Tree (Chap 2)
-			if ((tmem_addr + size) > TMEM_SIZE)
-				size = TMEM_SIZE - tmem_addr;
+				memcpy(texMem + tmem_addr_even, src_ptr, size);
+			}
+			else // RGBA8 tiles (and CI14, but that might just be stupid libogc!)
+			{
+				// AR and GB tiles are stored in separate TMEM banks => can't use a single memcpy for everything
+				u32 tmem_addr_odd = tmem_cfg.preload_tmem_odd * TMEM_LINE_SIZE;
 
-			memcpy(texMem + tmem_addr, ram_ptr, size);
+				for (u32 i = 0; i < tmem_cfg.preload_tile_info.count; ++i)
+				{
+					// FIXME: Duplicate conditions.
+					if (tmem_addr_even + TMEM_LINE_SIZE > TMEM_SIZE ||
+						tmem_addr_even + TMEM_LINE_SIZE > TMEM_SIZE)
+						break;
+
+					memcpy(texMem + tmem_addr_even, src_ptr, TMEM_LINE_SIZE);
+					memcpy(texMem + tmem_addr_odd, src_ptr + TMEM_LINE_SIZE, TMEM_LINE_SIZE);
+					tmem_addr_even += TMEM_LINE_SIZE;
+					tmem_addr_odd += TMEM_LINE_SIZE;
+					src_ptr += TMEM_LINE_SIZE * 2;
+				}
+			}
 		}
 		break;
 
diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
index 31aecfdc85..b3930c991c 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp
@@ -27,9 +27,8 @@
 #include "VideoConfig.h"
 #include "NativeVertexFormat.h"
 
-static int AlphaPreTest();
 
-static void StageHash(int stage, u32* out)
+static void StageHash(u32 stage, u32* out)
 {
 	out[0] |= bpmem.combiners[stage].colorC.hex & 0xFFFFFF; // 24
 	u32 alphaC = bpmem.combiners[stage].alphaC.hex & 0xFFFFF0; // 24, strip out tswap and rswap for now
@@ -109,29 +108,17 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	uid->values[0] |= bpmem.genMode.numtexgens << 4; // 4
 	uid->values[0] |= dstAlphaMode << 8; // 2
 
-	bool DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) || g_ActiveConfig.bEnablePerPixelDepth;
-
-	uid->values[0] |= DepthTextureEnable << 10; // 1
-
 	bool enablePL = g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting;
-	uid->values[0] |= enablePL << 11; // 1
+	uid->values[0] |= enablePL << 10; // 1
 
-	if (!enablePL) uid->values[0] |= xfregs.numTexGen.numTexGens << 12; // 4
-	u32 alphaPreTest = AlphaPreTest()+1;
+	if (!enablePL) uid->values[0] |= xfregs.numTexGen.numTexGens << 11; // 4
 
-	uid->values[0] |= alphaPreTest << 16; // 2
-
-	if (alphaPreTest == 1 || (alphaPreTest && !DepthTextureEnable && dstAlphaMode == DSTALPHA_ALPHA_PASS))
-	{
-		// Courtesy of PreAlphaTest, we're done already ;)
-		// NOTE: The comment header of generated shaders depends on the value of bpmem.genmode.numindstages.. shouldnt really bother about that though.
-		uid->num_values = 1;
-		return;
-	}
+	AlphaTest::TEST_RESULT alphaPreTest = bpmem.alpha_test.TestResult();
+	uid->values[0] |= alphaPreTest << 15; // 2
 
 	// numtexgens should be <= 8
 	for (unsigned int i = 0; i < bpmem.genMode.numtexgens; ++i)
-		uid->values[0] |= xfregs.texMtxInfo[i].projection << (18+i); // 1
+		uid->values[0] |= xfregs.texMtxInfo[i].projection << (17+i); // 1
 
 	uid->values[1] = bpmem.genMode.numindstages; // 3
 	u32 indirectStagesUsed = 0;
@@ -154,30 +141,24 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo
 	}
 
 	u32* ptr = &uid->values[2];
-	for (int i = 0; i < bpmem.genMode.numtevstages+1; ++i)
+	for (unsigned int i = 0; i < bpmem.genMode.numtevstages+1u; ++i)
 	{
 		StageHash(i, ptr);
 		ptr += 4; // max: ptr = &uid->values[66]
 	}
 
-	ptr[0] |= bpmem.alphaFunc.comp0; // 3
-	ptr[0] |= bpmem.alphaFunc.comp1 << 3; // 3
-	ptr[0] |= bpmem.alphaFunc.logic << 6; // 2
+	ptr[0] |= bpmem.alpha_test.comp0; // 3
+	ptr[0] |= bpmem.alpha_test.comp1 << 3; // 3
+	ptr[0] |= bpmem.alpha_test.logic << 6; // 2
 
-	if (alphaPreTest == 0 || alphaPreTest == 2)
-	{
-		ptr[0] |= bpmem.fog.c_proj_fsel.fsel << 8; // 3
-		if (DepthTextureEnable)
-		{
-			ptr[0] |= bpmem.ztex2.op << 11; // 2
-			ptr[0] |= bpmem.zcontrol.zcomploc << 13; // 1
-			ptr[0] |= bpmem.zmode.testenable << 14; // 1
-			ptr[0] |= bpmem.zmode.updateenable << 15; // 1
-		}
-	}
+	ptr[0] |= bpmem.ztex2.op << 8; // 2
+	ptr[0] |= bpmem.zcontrol.early_ztest << 10; // 1
+	ptr[0] |= bpmem.zmode.testenable << 11; // 1
+	ptr[0] |= bpmem.zmode.updateenable << 12; // 1
 
 	if (dstAlphaMode != DSTALPHA_ALPHA_PASS)
 	{
+		ptr[0] |= bpmem.fog.c_proj_fsel.fsel << 13; // 3
 		if (bpmem.fog.c_proj_fsel.fsel != 0)
 		{
 			ptr[0] |= bpmem.fog.c_proj_fsel.proj << 16; // 1
@@ -204,9 +185,8 @@ void GetSafePixelShaderId(PIXELSHADERUIDSAFE *uid, DSTALPHA_MODE dstAlphaMode, u
 	*ptr++ = bpmem.ztex2.hex; // 2
 	*ptr++ = bpmem.zcontrol.hex; // 3
 	*ptr++ = bpmem.zmode.hex; // 4
-	*ptr++ = g_ActiveConfig.bEnablePerPixelDepth; // 5
-	*ptr++ = g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting; // 6
-	*ptr++ = xfregs.numTexGen.hex; // 7
+	*ptr++ = g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting; // 5
+	*ptr++ = xfregs.numTexGen.hex; // 6
 
 	if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
 	{
@@ -218,28 +198,28 @@ void GetSafePixelShaderId(PIXELSHADERUIDSAFE *uid, DSTALPHA_MODE dstAlphaMode, u
 	}
 
 	for (unsigned int i = 0; i < 8; ++i)
-		*ptr++ = xfregs.texMtxInfo[i].hex; // 8-15
+		*ptr++ = xfregs.texMtxInfo[i].hex; // 7-14
 
 	for (unsigned int i = 0; i < 16; ++i)
-		*ptr++ = bpmem.tevind[i].hex; // 16-31
+		*ptr++ = bpmem.tevind[i].hex; // 15-30
 
-	*ptr++ = bpmem.tevindref.hex; // 32
+	*ptr++ = bpmem.tevindref.hex; // 31
 
-	for (u32 i = 0; i < bpmem.genMode.numtevstages+1; ++i) // up to 16 times
+	for (u32 i = 0; i < bpmem.genMode.numtevstages+1u; ++i) // up to 16 times
 	{
-		*ptr++ = bpmem.combiners[i].colorC.hex; // 33+5*i
-		*ptr++ = bpmem.combiners[i].alphaC.hex; // 34+5*i
-		*ptr++ = bpmem.tevind[i].hex; // 35+5*i
-		*ptr++ = bpmem.tevksel[i/2].hex; // 36+5*i
-		*ptr++ = bpmem.tevorders[i/2].hex; // 37+5*i
+		*ptr++ = bpmem.combiners[i].colorC.hex; // 32+5*i
+		*ptr++ = bpmem.combiners[i].alphaC.hex; // 33+5*i
+		*ptr++ = bpmem.tevind[i].hex; // 34+5*i
+		*ptr++ = bpmem.tevksel[i/2].hex; // 35+5*i
+		*ptr++ = bpmem.tevorders[i/2].hex; // 36+5*i
 	}
 
-	ptr = &uid->values[113];
+	ptr = &uid->values[112];
 
-	*ptr++ = bpmem.alphaFunc.hex; // 113
+	*ptr++ = bpmem.alpha_test.hex; // 112
 
-	*ptr++ = bpmem.fog.c_proj_fsel.hex; // 114
-	*ptr++ = bpmem.fogRange.Base.hex; // 115
+	*ptr++ = bpmem.fog.c_proj_fsel.hex; // 113
+	*ptr++ = bpmem.fogRange.Base.hex; // 114
 
 	_assert_((ptr - uid->values) == uid->GetNumValues());
 }
@@ -308,10 +288,10 @@ static const char *tevKSelTableC[] = // KCSEL
 	"0.375f,0.375f,0.375f", // 3_8 = 0x05
 	"0.25f,0.25f,0.25f",    // 1_4 = 0x06
 	"0.125f,0.125f,0.125f", // 1_8 = 0x07
-	"ERROR", // 0x08
-	"ERROR", // 0x09
-	"ERROR", // 0x0a
-	"ERROR", // 0x0b
+	"ERROR1", // 0x08
+	"ERROR2", // 0x09
+	"ERROR3", // 0x0a
+	"ERROR4", // 0x0b
 	I_KCOLORS"[0].rgb", // K0 = 0x0C
 	I_KCOLORS"[1].rgb", // K1 = 0x0D
 	I_KCOLORS"[2].rgb", // K2 = 0x0E
@@ -344,14 +324,14 @@ static const char *tevKSelTableA[] = // KASEL
 	"0.375f",// 3_8 = 0x05
 	"0.25f", // 1_4 = 0x06
 	"0.125f",// 1_8 = 0x07
-	"ERROR", // 0x08
-	"ERROR", // 0x09
-	"ERROR", // 0x0a
-	"ERROR", // 0x0b
-	"ERROR", // 0x0c
-	"ERROR", // 0x0d
-	"ERROR", // 0x0e
-	"ERROR", // 0x0f
+	"ERROR5", // 0x08
+	"ERROR6", // 0x09
+	"ERROR7", // 0x0a
+	"ERROR8", // 0x0b
+	"ERROR9", // 0x0c
+	"ERROR10", // 0x0d
+	"ERROR11", // 0x0e
+	"ERROR12", // 0x0f
 	I_KCOLORS"[0].r", // K0_R = 0x10
 	I_KCOLORS"[1].r", // K1_R = 0x11
 	I_KCOLORS"[2].r", // K2_R = 0x12
@@ -426,7 +406,7 @@ static const char *tevCInputTable[] = // CC
 	"float3(0.5f, 0.5f, 0.5f)",                 // HALF
 	"(ckonsttemp.rgb)", //"konsttemp.rgb",        // KONST
 	"float3(0.0f, 0.0f, 0.0f)",              // ZERO
-	"PADERROR", "PADERROR", "PADERROR", "PADERROR"
+	"PADERROR1", "PADERROR2", "PADERROR3", "PADERROR4"
 };
 
 static const char *tevAInputTable[] = // CA
@@ -448,17 +428,17 @@ static const char *tevAInputTable[] = // CA
 	"crastemp",         // RASA,
 	"ckonsttemp",       // KONST,  (hw1 had quarter)
 	"float4(0.0f, 0.0f, 0.0f, 0.0f)", // ZERO
-	"PADERROR", "PADERROR", "PADERROR", "PADERROR",
-	"PADERROR", "PADERROR", "PADERROR", "PADERROR",
+	"PADERROR5", "PADERROR6", "PADERROR7", "PADERROR8",
+	"PADERROR9", "PADERROR10", "PADERROR11", "PADERROR12",
 };
 
 static const char *tevRasTable[] =
 {
 	"colors_0",
 	"colors_1",
-	"ERROR", //2
-	"ERROR", //3
-	"ERROR", //4
+	"ERROR13", //2
+	"ERROR14", //3
+	"ERROR15", //4
 	"alphabump", // use bump alpha
 	"(alphabump*(255.0f/248.0f))", //normalized
 	"float4(0.0f, 0.0f, 0.0f, 0.0f)", // zero
@@ -481,7 +461,6 @@ static const char *tevIndFmtScale[]   = {"255.0f", "31.0f", "15.0f", "7.0f" };
 static char swapModeTable[4][5];
 
 static char text[16384];
-static bool DepthTextureEnable;
 
 struct RegisterState
 {
@@ -546,7 +525,6 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 				nIndirectStagesUsed |= 1 << bpmem.tevind[i].bt;
 		}
 	}
-	DepthTextureEnable = (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) || g_ActiveConfig.bEnablePerPixelDepth ;
 
 	if (ApiType == API_OPENGL)
 	{
@@ -639,8 +617,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
 			WRITE(p, "out float4 ocol1;\n");
 		
-		if (DepthTextureEnable)
-			WRITE(p, "float depth;\n");
+		WRITE(p, "float depth;\n");
 		WRITE(p, "float4 rawpos = gl_FragCoord;\n");
 
 		WRITE(p, "VARYIN float4 colors_02;\n");
@@ -695,14 +672,14 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		{
 			WRITE(p, "  out float4 ocol0 : COLOR0,%s%s\n  in float4 rawpos : %s,\n",
 				dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND ? "\n  out float4 ocol1 : COLOR1," : "",
-				DepthTextureEnable ? "\n  out float depth : DEPTH," : "",
+				"\n  out float depth : DEPTH,",
 				ApiType & API_D3D9_SM20 ? "POSITION" : "VPOS");
 		}
 		else
 		{
 			WRITE(p, "  out float4 ocol0 : SV_Target0,%s%s\n  in float4 rawpos : SV_Position,\n",
 				dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND ? "\n  out float4 ocol1 : SV_Target1," : "",
-				DepthTextureEnable ? "\n  out float depth : SV_Depth," : "");
+				"\n  out float depth : SV_Depth,");
 		}
 
 		WRITE(p, "  in float4 colors_0 : COLOR0,\n");
@@ -736,45 +713,14 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 		}
 	}
 
-	int Pretest = AlphaPreTest();
-	if(Pretest >= 0 && !DepthTextureEnable)
-	{
-		if (!Pretest)
-		{
-			// alpha test will always fail, so restart the shader and just make it an empty function		
-			WRITE(p, "\tocol0 = float4(0.0f, 0.0f, 0.0f, 0.0f);\n");
-			if(DepthTextureEnable)
-				WRITE(p, "\tdepth = 1.f;\n");
-			if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
-				WRITE(p, "\tocol1 = float4(0.0f, 0.0f, 0.0f, 0.0f);\n");
-			if (ApiType == API_OPENGL)
-			{
-				if (DepthTextureEnable)
-					WRITE(p, "\tgl_FragDepth = depth;\n");
-			}
-			WRITE(p, "\tdiscard;\n");
-			if(ApiType != API_D3D11)
-				WRITE(p, "\treturn;\n");
-		}
-		else if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
-		{
-			WRITE(p, "\tocol0 = " I_ALPHA"[0].aaaa;\n");
-		}
-		if(!Pretest || dstAlphaMode == DSTALPHA_ALPHA_PASS)
-		{
-			WRITE(p, "}\n");
-			return text;
-		}
-	}
-
-	WRITE(p, "\tfloat4 c0 = " I_COLORS"[1], c1 = " I_COLORS"[2], c2 = " I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n"
-			"\tfloat3 comp16 = float3(1.0f, 255.0f, 0.0f), comp24 = float3(1.0f, 255.0f, 255.0f*255.0f);\n"
-			"\tfloat4 alphabump=float4(0.0f,0.0f,0.0f,0.0f);\n"
-			"\tfloat3 tevcoord=float3(0.0f, 0.0f, 0.0f);\n"
-			"\tfloat2 wrappedcoord=float2(0.0f,0.0f), tempcoord=float2(0.0f,0.0f);\n"
-			"\tfloat4 cc0=float4(0.0f,0.0f,0.0f,0.0f), cc1=float4(0.0f,0.0f,0.0f,0.0f);\n"
-			"\tfloat4 cc2=float4(0.0f,0.0f,0.0f,0.0f), cprev=float4(0.0f,0.0f,0.0f,0.0f);\n"
-			"\tfloat4 crastemp=float4(0.0f,0.0f,0.0f,0.0f),ckonsttemp=float4(0.0f,0.0f,0.0f,0.0f);\n\n");
+	WRITE(p, "  float4 c0 = " I_COLORS"[1], c1 = " I_COLORS"[2], c2 = " I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n"
+			"  float3 comp16 = float3(1.0f, 255.0f, 0.0f), comp24 = float3(1.0f, 255.0f, 255.0f*255.0f);\n"
+			"  float4 alphabump=float4(0.0f,0.0f,0.0f,0.0f);\n"
+			"  float3 tevcoord=float3(0.0f, 0.0f, 0.0f);\n"
+			"  float2 wrappedcoord=float2(0.0f,0.0f), tempcoord=float2(0.0f,0.0f);\n"
+			"  float4 cc0=float4(0.0f,0.0f,0.0f,0.0f), cc1=float4(0.0f,0.0f,0.0f,0.0f);\n"
+			"  float4 cc2=float4(0.0f,0.0f,0.0f,0.0f), cprev=float4(0.0f,0.0f,0.0f,0.0f);\n"
+			"  float4 crastemp=float4(0.0f,0.0f,0.0f,0.0f),ckonsttemp=float4(0.0f,0.0f,0.0f,0.0f);\n\n");
 
 	if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
 	{
@@ -874,34 +820,26 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	if(RegisterStates[0].AlphaNeedOverflowControl || RegisterStates[0].ColorNeedOverflowControl)
 		WRITE(p, "\tprev = frac(prev * (255.0f/256.0f)) * (256.0f/255.0f);\n");
 
-	if(Pretest ==  -1)
-	{
+	AlphaTest::TEST_RESULT Pretest = bpmem.alpha_test.TestResult();
+	if (Pretest == AlphaTest::UNDETERMINED)
 		WriteAlphaTest(p, ApiType, dstAlphaMode);
-	}
 
-	if((bpmem.fog.c_proj_fsel.fsel != 0) || DepthTextureEnable)
-	{
-		// the screen space depth value = far z + (clip z / clip w) * z range
-		WRITE(p, "\tfloat zCoord = " I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * " I_ZBIAS"[1].y;\n");
-	}
+	// the screen space depth value = far z + (clip z / clip w) * z range
+	WRITE(p, "float zCoord = " I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * " I_ZBIAS"[1].y;\n");
 
-	if (DepthTextureEnable)
+	// Note: depth textures are disabled if early depth test is enabled
+	if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.early_ztest && bpmem.zmode.testenable)
 	{
 		// use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format...
-		if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable)
-		{
-			if (bpmem.ztex2.op == ZTEXTURE_ADD)
-				WRITE(p, "\tzCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w + zCoord;\n");
-			else
-				WRITE(p, "\tzCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w;\n");
+		WRITE(p, "zCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w %s;\n",
+									(bpmem.ztex2.op == ZTEXTURE_ADD) ? "+ zCoord" : "");
 
-			// scale to make result from frac correct
-			WRITE(p, "\tzCoord = zCoord * (16777215.0f/16777216.0f);\n");
-			WRITE(p, "\tzCoord = frac(zCoord);\n");
-			WRITE(p, "\tzCoord = zCoord * (16777216.0f/16777215.0f);\n");
-		}
-		WRITE(p, "\tdepth = zCoord;\n");
+		// scale to make result from frac correct
+		WRITE(p, "zCoord = zCoord * (16777215.0f/16777216.0f);\n");
+		WRITE(p, "zCoord = frac(zCoord);\n");
+		WRITE(p, "zCoord = zCoord * (16777216.0f/16777215.0f);\n");
 	}
+	WRITE(p, "depth = zCoord;\n");
 
 	if (dstAlphaMode == DSTALPHA_ALPHA_PASS)
 		WRITE(p, "\tocol0 = float4(prev.rgb, " I_ALPHA"[0].a);\n");
@@ -923,8 +861,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType
 	
 	if (ApiType == API_OPENGL)
 	{
-		if (DepthTextureEnable)
-			WRITE(p, "\tgl_FragDepth = depth;\n");
+		WRITE(p, "\tgl_FragDepth = depth;\n");
 	}
 	WRITE(p, "}\n");
 	if (text[sizeof(text) - 1] != 0x7C)
@@ -1316,39 +1253,6 @@ static const char *tevAlphaFunclogicTable[] =
 	" != ", // xor
 	" == "  // xnor
 };
-static int AlphaPreTest()
-{
-	u32 op = bpmem.alphaFunc.logic;
-	u32 comp[2] = {bpmem.alphaFunc.comp0, bpmem.alphaFunc.comp1};
-
-	// First kill all the simple cases
-	switch(op)
-	{
-	case 0: // AND
-		if (comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_ALWAYS) return true;
-		if (comp[0] == ALPHACMP_NEVER || comp[1] == ALPHACMP_NEVER) return false;
-		break;
-	case 1: // OR
-		if (comp[0] == ALPHACMP_ALWAYS || comp[1] == ALPHACMP_ALWAYS) return true;
-		if (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_NEVER)return false;
-		break;
-	case 2: // XOR
-		if ((comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_NEVER) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_ALWAYS))
-			return true;
-		if ((comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_ALWAYS) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_NEVER))
-			return false;
-		break;
-	case 3: // XNOR
-		if ((comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_NEVER) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_ALWAYS))
-			return false;
-		if ((comp[0] == ALPHACMP_ALWAYS && comp[1] == ALPHACMP_ALWAYS) || (comp[0] == ALPHACMP_NEVER && comp[1] == ALPHACMP_NEVER))
-			return true;
-		break;
-	default: PanicAlert("bad logic for alpha test? %08x", op);
-	}
-	return -1;
-}
-
 
 static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode)
 {
@@ -1362,41 +1266,38 @@ static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode
 	// using discard then return works the same in cg and dx9 but not in dx11
 	WRITE(p, "\tif(!( ");
 
-	int compindex = bpmem.alphaFunc.comp0 % 8;
+	int compindex = bpmem.alpha_test.comp0;
 	WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[0]);//lookup the first component from the alpha function table
 
-	WRITE(p, "%s", tevAlphaFunclogicTable[bpmem.alphaFunc.logic % 4]);//lookup the logic op
+	WRITE(p, "%s", tevAlphaFunclogicTable[bpmem.alpha_test.logic]);//lookup the logic op
 
-	compindex = bpmem.alphaFunc.comp1 % 8;
+	compindex = bpmem.alpha_test.comp1;
 	WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[1]);//lookup the second component from the alpha function table
 	WRITE(p, ")) {\n");
 
 	WRITE(p, "\t\tocol0 = float4(0.0f, 0.0f, 0.0f, 0.0f);\n");
 	if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND)
 		WRITE(p, "\t\tocol1 = float4(0.0f, 0.0f, 0.0f, 0.0f);\n");
-	if (DepthTextureEnable)
-		WRITE(p, "\t\tdepth = 1.f;\n");
-
-	// HAXX: zcomploc is a way to control whether depth test is done before
-	// or after texturing and alpha test. PC GPU does depth test before texturing ONLY if depth value is
-	// not updated during shader execution.
+	WRITE(p, "depth = 1.f;\n");
+
+	// HAXX: zcomploc (aka early_ztest) is a way to control whether depth test is done before
+	// or after texturing and alpha test. PC GPUs have no way to support this
+	// feature properly as of 2012: depth buffer and depth test are not
+	// programmable and the depth test is always done after texturing.
+	// Most importantly, PC GPUs do not allow writing to the z buffer without
+	// writing a color value (unless color writing is disabled altogether).
 	// We implement "depth test before texturing" by discarding the fragment
 	// when the alpha test fail. This is not a correct implementation because
-	// even if the depth test fails the fragment could be alpha blended.
-	// this implemnetation is a trick to  keep speed.
-	// the correct, but slow, way to implement a correct zComploc is :
-	// 1 - if zcomplock is enebled make a first pass, with color channel write disabled updating only 
-	// depth channel.
-	// 2 - in the next pass disable depth chanel update, but proccess the color data normally
-	// this way is the only CORRECT way to emulate perfectly the zcomplock behaviour
-	if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable))
+	// even if the depth test fails the fragment could be alpha blended, but
+	// we don't have a choice.
+	if (!(bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable))
 	{
 		WRITE(p, "\t\tdiscard;\n");
 		if (ApiType != API_D3D11)
 			WRITE(p, "\t\treturn;\n");
 	}
 
-	WRITE(p, "\t}\n");
+	WRITE(p, "}\n");
 }
 
 static const char *tevFogFuncsTable[] =
diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.h b/Source/Core/VideoCommon/Src/PixelShaderGen.h
index 1a70a3664b..03ddd6d001 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderGen.h
+++ b/Source/Core/VideoCommon/Src/PixelShaderGen.h
@@ -45,7 +45,7 @@
 #define C_PMATERIALS	(C_PLIGHTS + 40)
 #define C_PENVCONST_END (C_PMATERIALS + 4)
 #define PIXELSHADERUID_MAX_VALUES 70
-#define PIXELSHADERUID_MAX_VALUES_SAFE 120
+#define PIXELSHADERUID_MAX_VALUES_SAFE 115
 
 // Annoying sure, can be removed once we get up to GLSL ~1.3
 const s_svar PSVar_Loc[] = { {I_COLORS, C_COLORS, 4 },
diff --git a/Source/Core/VideoCommon/Src/PixelShaderManager.cpp b/Source/Core/VideoCommon/Src/PixelShaderManager.cpp
index 651f0c73dd..f744f41ee1 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderManager.cpp
+++ b/Source/Core/VideoCommon/Src/PixelShaderManager.cpp
@@ -375,7 +375,7 @@ void PixelShaderManager::SetColorChanged(int type, int num, bool high)
     PRIM_LOG("pixel %scolor%d: %f %f %f %f\n", type?"k":"", num, pf[0], pf[1], pf[2], pf[3]);
 }
 
-void PixelShaderManager::SetAlpha(const AlphaFunc& alpha)
+void PixelShaderManager::SetAlpha(const AlphaTest& alpha)
 {
     if ((alpha.hex & 0xffff) != lastAlpha)
 	{
diff --git a/Source/Core/VideoCommon/Src/PixelShaderManager.h b/Source/Core/VideoCommon/Src/PixelShaderManager.h
index a336287a6f..12d749c871 100644
--- a/Source/Core/VideoCommon/Src/PixelShaderManager.h
+++ b/Source/Core/VideoCommon/Src/PixelShaderManager.h
@@ -38,7 +38,7 @@ public:
 
 	// constant management, should be called after memory is committed
 	static void SetColorChanged(int type, int index, bool high);
-	static void SetAlpha(const AlphaFunc& alpha);
+	static void SetAlpha(const AlphaTest& alpha);
 	static void SetDestAlpha(const ConstantAlpha& alpha);
 	static void SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt);
 	static void SetZTextureBias(u32 bias);
diff --git a/Source/Core/VideoCommon/Src/RenderBase.cpp b/Source/Core/VideoCommon/Src/RenderBase.cpp
index 577570c568..8e1d012fa2 100644
--- a/Source/Core/VideoCommon/Src/RenderBase.cpp
+++ b/Source/Core/VideoCommon/Src/RenderBase.cpp
@@ -91,6 +91,9 @@ Renderer::Renderer() : frame_data(NULL), bLastFrameDumped(false)
 #if defined _WIN32 || defined HAVE_LIBAV
 	bAVIDumping = false;
 #endif
+
+	OSDChoice = 0;
+	OSDTime = 0;
 }
 
 Renderer::~Renderer()
@@ -263,108 +266,109 @@ void Renderer::SetScreenshot(const char *filename)
 // Create On-Screen-Messages
 void Renderer::DrawDebugText()
 {
+	if (!g_Config.bOSDHotKey)
+		return;
+
 	// OSD Menu messages
-	if (g_ActiveConfig.bOSDHotKey)
+	if (OSDChoice > 0)
 	{
-		if (OSDChoice > 0)
-		{
-			OSDTime = Common::Timer::GetTimeMs() + 3000;
-			OSDChoice = -OSDChoice;
-		}
-		if ((u32)OSDTime > Common::Timer::GetTimeMs())
-		{
-			const char* res_text = "";
-			switch (g_ActiveConfig.iEFBScale)
-			{
-			case 0:
-				res_text = "Auto (fractional)";
-				break;
-			case 1:
-				res_text = "Auto (integral)";
-				break;
-			case 2:
-				res_text = "Native";
-				break;
-			case 3:
-				res_text = "1.5x";
-				break;
-			case 4:
-				res_text = "2x";
-				break;
-			case 5:
-				res_text = "2.5x";
-				break;
-			case 6:
-				res_text = "3x";
-				break;
-			case 7:
-				res_text = "4x";
-				break;
-			}
+		OSDTime = Common::Timer::GetTimeMs() + 3000;
+		OSDChoice = -OSDChoice;
+	}
 
-			const char* ar_text = "";
-			switch(g_ActiveConfig.iAspectRatio)
-			{
-			case ASPECT_AUTO:
-				ar_text = "Auto";
-				break;
-			case ASPECT_FORCE_16_9:
-				ar_text = "16:9";
-				break;
-			case ASPECT_FORCE_4_3:
-				ar_text = "4:3";
-				break;
-			case ASPECT_STRETCH:
-				ar_text = "Stretch";
-				break;
-			}
+	if ((u32)OSDTime <= Common::Timer::GetTimeMs())
+		return;
 
-			const char* const efbcopy_text = g_ActiveConfig.bEFBCopyEnable ?
-				(g_ActiveConfig.bCopyEFBToTexture ? "to Texture" : "to RAM") : "Disabled";
+	const char* res_text = "";
+	switch (g_ActiveConfig.iEFBScale)
+	{
+	case 0:
+		res_text = "Auto (fractional)";
+		break;
+	case 1:
+		res_text = "Auto (integral)";
+		break;
+	case 2:
+		res_text = "Native";
+		break;
+	case 3:
+		res_text = "1.5x";
+		break;
+	case 4:
+		res_text = "2x";
+		break;
+	case 5:
+		res_text = "2.5x";
+		break;
+	case 6:
+		res_text = "3x";
+		break;
+	case 7:
+		res_text = "4x";
+		break;
+	}
 
-			// The rows
-			const std::string lines[] =
-			{
-				std::string("3: Internal Resolution: ") + res_text,
-				std::string("4: Aspect Ratio: ") + ar_text + (g_ActiveConfig.bCrop ? " (crop)" : ""),
-				std::string("5: Copy EFB: ") + efbcopy_text,
-				std::string("6: Fog: ") + (g_ActiveConfig.bDisableFog ? "Disabled" : "Enabled"),
-			};
+	const char* ar_text = "";
+	switch(g_ActiveConfig.iAspectRatio)
+	{
+	case ASPECT_AUTO:
+		ar_text = "Auto";
+		break;
+	case ASPECT_FORCE_16_9:
+		ar_text = "16:9";
+		break;
+	case ASPECT_FORCE_4_3:
+		ar_text = "4:3";
+		break;
+	case ASPECT_STRETCH:
+		ar_text = "Stretch";
+		break;
+	}
 
-			enum { lines_count = sizeof(lines)/sizeof(*lines) };
+	const char* const efbcopy_text = g_ActiveConfig.bEFBCopyEnable ?
+		(g_ActiveConfig.bCopyEFBToTexture ? "to Texture" : "to RAM") : "Disabled";
 
-			std::string final_yellow, final_cyan;
+	// The rows
+	const std::string lines[] =
+	{
+		std::string("3: Internal Resolution: ") + res_text,
+		std::string("4: Aspect Ratio: ") + ar_text + (g_ActiveConfig.bCrop ? " (crop)" : ""),
+		std::string("5: Copy EFB: ") + efbcopy_text,
+		std::string("6: Fog: ") + (g_ActiveConfig.bDisableFog ? "Disabled" : "Enabled"),
+	};
 
-			// If there is more text than this we will have a collision
-			if (g_ActiveConfig.bShowFPS)
-			{
-				final_yellow = final_cyan = "\n\n";
-			}
+	enum { lines_count = sizeof(lines)/sizeof(*lines) };
 
-			// The latest changed setting in yellow
-			for (int i = 0; i != lines_count; ++i)
-			{
-				if (OSDChoice == -i - 1)
-					final_yellow += lines[i];
-				final_yellow += '\n';
-			}
+	std::string final_yellow, final_cyan;
 
-			// The other settings in cyan
-			for (int i = 0; i != lines_count; ++i)
-			{
-				if (OSDChoice != -i - 1)
-					final_cyan += lines[i];
-				final_cyan += '\n';
-			}
+	// If there is more text than this we will have a collision
+	if (g_ActiveConfig.bShowFPS)
+	{
+		final_yellow = final_cyan = "\n\n";
+	}
 
-			// Render a shadow
-			g_renderer->RenderText(final_cyan.c_str(), 21, 21, 0xDD000000);
-			g_renderer->RenderText(final_yellow.c_str(), 21, 21, 0xDD000000);
-			//and then the text
-			g_renderer->RenderText(final_cyan.c_str(), 20, 20, 0xFF00FFFF);
-			g_renderer->RenderText(final_yellow.c_str(), 20, 20, 0xFFFFFF00);
-		}
+	// The latest changed setting in yellow
+	for (int i = 0; i != lines_count; ++i)
+	{
+		if (OSDChoice == -i - 1)
+			final_yellow += lines[i];
+		final_yellow += '\n';
 	}
+
+	// The other settings in cyan
+	for (int i = 0; i != lines_count; ++i)
+	{
+		if (OSDChoice != -i - 1)
+			final_cyan += lines[i];
+		final_cyan += '\n';
+	}
+
+	// Render a shadow
+	g_renderer->RenderText(final_cyan.c_str(), 21, 21, 0xDD000000);
+	g_renderer->RenderText(final_yellow.c_str(), 21, 21, 0xDD000000);
+	//and then the text
+	g_renderer->RenderText(final_cyan.c_str(), 20, 20, 0xFF00FFFF);
+	g_renderer->RenderText(final_yellow.c_str(), 20, 20, 0xFFFFFF00);
 }
 
 // TODO: remove
diff --git a/Source/Core/VideoCommon/Src/RenderBase.h b/Source/Core/VideoCommon/Src/RenderBase.h
index 4da80c9752..5376577cbd 100644
--- a/Source/Core/VideoCommon/Src/RenderBase.h
+++ b/Source/Core/VideoCommon/Src/RenderBase.h
@@ -39,7 +39,7 @@
 
 // TODO: Move these out of here.
 extern int frameCount;
-extern int OSDChoice, OSDTime;
+extern int OSDChoice;
 
 extern bool bLastFrameDumped;
 
diff --git a/Source/Core/VideoCommon/Src/Statistics.h b/Source/Core/VideoCommon/Src/Statistics.h
index 7238398b42..e173bfc682 100644
--- a/Source/Core/VideoCommon/Src/Statistics.h
+++ b/Source/Core/VideoCommon/Src/Statistics.h
@@ -43,7 +43,7 @@ struct Statistics
 
 	int numUniquePixelShaders;
 
-	float proj_0, proj_1, proj_2, proj_3, proj_4, proj_5, proj_6;
+	float proj_0, proj_1, proj_2, proj_3, proj_4, proj_5;
 	float gproj_0, gproj_1, gproj_2, gproj_3, gproj_4, gproj_5;
 	float gproj_6, gproj_7, gproj_8, gproj_9, gproj_10, gproj_11, gproj_12, gproj_13, gproj_14, gproj_15;
 
diff --git a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
index c2f1466be7..58b2fac341 100644
--- a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
+++ b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp
@@ -204,12 +204,16 @@ void TextureCache::ClearRenderTargets()
 		iter = textures.begin(),
 		tcend = textures.end();
 
-	for (; iter!=tcend; ++iter)
+	while (iter != tcend)
+	{
 		if (iter->second->type == TCET_EC_VRAM)
 		{
 			delete iter->second;
 			textures.erase(iter++);
 		}
+		else
+			++iter;
+	}
 }
 
 bool TextureCache::CheckForCustomTextureLODs(u64 tex_hash, int texformat, unsigned int levels)
@@ -329,6 +333,7 @@ TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage,
 	if (from_tmem) src_data = &texMem[bpmem.tex[stage/4].texImage1[stage%4].tmem_even * TMEM_LINE_SIZE];
 	else src_data = Memory::GetPointer(address);
 
+	// TODO: This doesn't hash GB tiles for preloaded RGBA8 textures (instead, it's hashing more data from the low tmem bank than it should)
 	tex_hash = GetHash64(src_data, texture_size, g_ActiveConfig.iSafeTextureCache_ColorSamples);
 	if (isPaletteTexture)
 	{
@@ -409,10 +414,19 @@ TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage,
 		}
 	}
 
-	// TODO: RGBA8 textures are stored non-continuously in tmem, that might cause problems here when preloading is enabled
 	if (!using_custom_texture)
-		pcfmt = TexDecoder_Decode(temp, src_data, expandedWidth,
-					expandedHeight, texformat, tlutaddr, tlutfmt, g_ActiveConfig.backend_info.bUseRGBATextures);
+	{
+		if (!(texformat == GX_TF_RGBA8 && from_tmem))
+		{
+			pcfmt = TexDecoder_Decode(temp, src_data, expandedWidth,
+						expandedHeight, texformat, tlutaddr, tlutfmt, g_ActiveConfig.backend_info.bUseRGBATextures);
+		}
+		else
+		{
+			u8* src_data_gb = &texMem[bpmem.tex[stage/4].texImage2[stage%4].tmem_odd * TMEM_LINE_SIZE];
+			pcfmt = TexDecoder_DecodeRGBA8FromTmem(temp, src_data, src_data_gb, expandedWidth, expandedHeight);
+		}
+	}
 
 	// TODO: Cleanup. Plus, we still autogenerate mipmaps in certain cases (we shouldn't do that)
 	bool isPow2;
diff --git a/Source/Core/VideoCommon/Src/TextureDecoder.cpp b/Source/Core/VideoCommon/Src/TextureDecoder.cpp
index 180c901730..44efcf8bae 100644
--- a/Source/Core/VideoCommon/Src/TextureDecoder.cpp
+++ b/Source/Core/VideoCommon/Src/TextureDecoder.cpp
@@ -2515,6 +2515,40 @@ void TexDecoder_DecodeTexel(u8 *dst, const u8 *src, int s, int t, int imageWidth
 	}
 }
 
+void TexDecoder_DecodeTexelRGBA8FromTmem(u8 *dst, const u8 *src_ar, const u8* src_gb, int s, int t, int imageWidth)
+{
+	u16 sBlk = s >> 2;
+	u16 tBlk = t >> 2;
+	u16 widthBlks = (imageWidth >> 2) + 1; // TODO: Looks wrong. Shouldn't this be ((imageWidth-1)>>2)+1 ?
+	u32 base_ar = (tBlk * widthBlks + sBlk) << 4;
+	u32 base_gb = (tBlk * widthBlks + sBlk) << 4;
+	u16 blkS = s & 3;
+	u16 blkT =  t & 3;
+	u32 blk_off = (blkT << 2) + blkS;
+
+	u32 offset_ar = (base_ar + blk_off) << 1;
+	u32 offset_gb = (base_gb + blk_off) << 1;
+	const u8* val_addr_ar = src_ar + offset_ar;
+	const u8* val_addr_gb = src_gb + offset_gb;
+
+	dst[3] = val_addr_ar[0]; // A
+	dst[0] = val_addr_ar[1]; // R
+	dst[1] = val_addr_gb[0]; // G
+	dst[2] = val_addr_gb[1]; // B
+}
+
+PC_TexFormat TexDecoder_DecodeRGBA8FromTmem(u8* dst, const u8 *src_ar, const u8 *src_gb, int width, int height)
+{
+	// TODO for someone who cares: Make this less slow!
+	for (int y = 0; y < height; ++y)
+		for (int x = 0; x < width; ++x)
+		{
+			TexDecoder_DecodeTexelRGBA8FromTmem(dst, src_ar, src_gb, x, y, width-1);
+			dst += 4;
+		}
+
+	return PC_TEX_FMT_RGBA32;
+}
 
 const char* texfmt[] = {
 	// pixel
diff --git a/Source/Core/VideoCommon/Src/TextureDecoder.h b/Source/Core/VideoCommon/Src/TextureDecoder.h
index 1901c86d27..fae70897f5 100644
--- a/Source/Core/VideoCommon/Src/TextureDecoder.h
+++ b/Source/Core/VideoCommon/Src/TextureDecoder.h
@@ -86,6 +86,8 @@ enum PC_TexFormat
 PC_TexFormat TexDecoder_Decode(u8 *dst, const u8 *src, int width, int height, int texformat, int tlutaddr, int tlutfmt,bool rgbaOnly = false);
 PC_TexFormat GetPC_TexFormat(int texformat, int tlutfmt);
 void TexDecoder_DecodeTexel(u8 *dst, const u8 *src, int s, int t, int imageWidth, int texformat, int tlutaddr, int tlutfmt);
+void TexDecoder_DecodeTexelRGBA8FromTmem(u8 *dst, const u8 *src_ar, const u8* src_gb, int s, int t, int imageWidth);
+PC_TexFormat TexDecoder_DecodeRGBA8FromTmem(u8* dst, const u8 *src_ar, const u8 *src_gb, int width, int height);
 void TexDecoder_SetTexFmtOverlayOptions(bool enable, bool center);
 
 #endif
diff --git a/Source/Core/VideoCommon/Src/VertexManagerBase.cpp b/Source/Core/VideoCommon/Src/VertexManagerBase.cpp
index bcd46c5faf..bae5711e14 100644
--- a/Source/Core/VideoCommon/Src/VertexManagerBase.cpp
+++ b/Source/Core/VideoCommon/Src/VertexManagerBase.cpp
@@ -205,7 +205,7 @@ void VertexManager::Flush()
 	}
 
 	PRIM_LOG("pixel: tev=%d, ind=%d, texgen=%d, dstalpha=%d, alphafunc=0x%x", bpmem.genMode.numtevstages+1, bpmem.genMode.numindstages,
-		bpmem.genMode.numtexgens, (u32)bpmem.dstalpha.enable, (bpmem.alphaFunc.hex>>16)&0xff);
+		bpmem.genMode.numtexgens, (u32)bpmem.dstalpha.enable, (bpmem.alpha_test.hex>>16)&0xff);
 #endif
 
 	u32 usedtextures = 0;
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
index 260d2de5e5..dc66041643 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
@@ -542,6 +542,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE ApiType)
 
 	//write the true depth value, if the game uses depth textures pixel shaders will override with the correct values
 	//if not early z culling will improve speed
+	// TODO: Can probably be dropped?
 	if (is_d3d)
 	{
 		WRITE(p, "o.pos.z = " I_DEPTHPARAMS".x * o.pos.w + o.pos.z * " I_DEPTHPARAMS".y;\n");
diff --git a/Source/Core/VideoCommon/Src/VertexShaderManager.cpp b/Source/Core/VideoCommon/Src/VertexShaderManager.cpp
index f21ae396e9..b3a7db8af2 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderManager.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderManager.cpp
@@ -169,14 +169,24 @@ void VertexShaderManager::Shutdown()
 
 void VertexShaderManager::Dirty()
 {
-	nTransformMatricesChanged[0] = 0; nTransformMatricesChanged[1] = 256;
-	nNormalMatricesChanged[0] = 0; nNormalMatricesChanged[1] = 96;
-	nPostTransformMatricesChanged[0] = 0; nPostTransformMatricesChanged[1] = 256;
-	nLightsChanged[0] = 0; nLightsChanged[1] = 0x80;
+	nTransformMatricesChanged[0] = 0; 
+	nTransformMatricesChanged[1] = 256;
+	
+	nNormalMatricesChanged[0] = 0;
+	nNormalMatricesChanged[1] = 96;
+	
+	nPostTransformMatricesChanged[0] = 0; 
+	nPostTransformMatricesChanged[1] = 256;
+	
+	nLightsChanged[0] = 0; 
+	nLightsChanged[1] = 0x80;
+	
 	bPosNormalMatrixChanged = true;
-	bTexMatricesChanged[0] = bTexMatricesChanged[1] = true;
+	bTexMatricesChanged[0] = true;
+	bTexMatricesChanged[1] = true;
+	
 	bProjectionChanged = true;
-	bPosNormalMatrixChanged = bTexMatricesChanged[0] = bTexMatricesChanged[1] = true;
+	
 	nMaterialsChanged = 15;
 }
 
@@ -342,26 +352,28 @@ void VertexShaderManager::SetConstants()
 	if (bProjectionChanged)
 	{
 		bProjectionChanged = false;
+		
+		float *rawProjection = xfregs.projection.rawProjection;
 
-		if (xfregs.rawProjection[6] == 0)
+		switch(xfregs.projection.type)
 		{
-			// Perspective
+		case GX_PERSPECTIVE:
 			
-			g_fProjectionMatrix[0] = xfregs.rawProjection[0] * g_ActiveConfig.fAspectRatioHackW;
+			g_fProjectionMatrix[0] = rawProjection[0] * g_ActiveConfig.fAspectRatioHackW;
 			g_fProjectionMatrix[1] = 0.0f;
-			g_fProjectionMatrix[2] = xfregs.rawProjection[1];
+			g_fProjectionMatrix[2] = rawProjection[1];
 			g_fProjectionMatrix[3] = 0.0f;
 
 			g_fProjectionMatrix[4] = 0.0f;
-			g_fProjectionMatrix[5] = xfregs.rawProjection[2] * g_ActiveConfig.fAspectRatioHackH;
-			g_fProjectionMatrix[6] = xfregs.rawProjection[3];
+			g_fProjectionMatrix[5] = rawProjection[2] * g_ActiveConfig.fAspectRatioHackH;
+			g_fProjectionMatrix[6] = rawProjection[3];
 			g_fProjectionMatrix[7] = 0.0f;
 
 			g_fProjectionMatrix[8] = 0.0f;
 			g_fProjectionMatrix[9] = 0.0f;
-			g_fProjectionMatrix[10] = xfregs.rawProjection[4];
+			g_fProjectionMatrix[10] = rawProjection[4];
 
-			g_fProjectionMatrix[11] = xfregs.rawProjection[5];
+			g_fProjectionMatrix[11] = rawProjection[5];
  			
 			g_fProjectionMatrix[12] = 0.0f;
 			g_fProjectionMatrix[13] = 0.0f;
@@ -386,24 +398,24 @@ void VertexShaderManager::SetConstants()
 			SETSTAT_FT(stats.gproj_13, g_fProjectionMatrix[13]);
 			SETSTAT_FT(stats.gproj_14, g_fProjectionMatrix[14]);
 			SETSTAT_FT(stats.gproj_15, g_fProjectionMatrix[15]);
-		}
-		else
-		{ 
-			// Orthographic Projection
-			g_fProjectionMatrix[0] = xfregs.rawProjection[0];
+			break;
+			
+		case GX_ORTHOGRAPHIC:
+			
+			g_fProjectionMatrix[0] = rawProjection[0];
 			g_fProjectionMatrix[1] = 0.0f;
 			g_fProjectionMatrix[2] = 0.0f;
-			g_fProjectionMatrix[3] = xfregs.rawProjection[1];
+			g_fProjectionMatrix[3] = rawProjection[1];
 
 			g_fProjectionMatrix[4] = 0.0f;
-			g_fProjectionMatrix[5] = xfregs.rawProjection[2];
+			g_fProjectionMatrix[5] = rawProjection[2];
 			g_fProjectionMatrix[6] = 0.0f;
-			g_fProjectionMatrix[7] = xfregs.rawProjection[3];
+			g_fProjectionMatrix[7] = rawProjection[3];
 
 			g_fProjectionMatrix[8] = 0.0f;
 			g_fProjectionMatrix[9] = 0.0f;
-			g_fProjectionMatrix[10] = (g_ProjHack1.value + xfregs.rawProjection[4]) * ((g_ProjHack1.sign == 0) ? 1.0f : g_ProjHack1.sign);
-			g_fProjectionMatrix[11] = (g_ProjHack2.value + xfregs.rawProjection[5]) * ((g_ProjHack2.sign == 0) ? 1.0f : g_ProjHack2.sign);
+			g_fProjectionMatrix[10] = (g_ProjHack1.value + rawProjection[4]) * ((g_ProjHack1.sign == 0) ? 1.0f : g_ProjHack1.sign);
+			g_fProjectionMatrix[11] = (g_ProjHack2.value + rawProjection[5]) * ((g_ProjHack2.sign == 0) ? 1.0f : g_ProjHack2.sign);
 
 			g_fProjectionMatrix[12] = 0.0f;
 			g_fProjectionMatrix[13] = 0.0f;
@@ -416,7 +428,7 @@ void VertexShaderManager::SetConstants()
 			*/
 			
 			g_fProjectionMatrix[14] = 0.0f;
-			g_fProjectionMatrix[15] = (g_ProjHack3 && xfregs.rawProjection[0] == 2.0f ? 0.0f : 1.0f);  //causes either the efb copy or bloom layer not to show if proj hack enabled
+			g_fProjectionMatrix[15] = (g_ProjHack3 && rawProjection[0] == 2.0f ? 0.0f : 1.0f);  //causes either the efb copy or bloom layer not to show if proj hack enabled
 		
 			SETSTAT_FT(stats.g2proj_0, g_fProjectionMatrix[0]);
 			SETSTAT_FT(stats.g2proj_1, g_fProjectionMatrix[1]);
@@ -434,18 +446,21 @@ void VertexShaderManager::SetConstants()
 			SETSTAT_FT(stats.g2proj_13, g_fProjectionMatrix[13]);
 			SETSTAT_FT(stats.g2proj_14, g_fProjectionMatrix[14]);
 			SETSTAT_FT(stats.g2proj_15, g_fProjectionMatrix[15]);
-			SETSTAT_FT(stats.proj_0, xfregs.rawProjection[0]);
-			SETSTAT_FT(stats.proj_1, xfregs.rawProjection[1]);
-			SETSTAT_FT(stats.proj_2, xfregs.rawProjection[2]);
-			SETSTAT_FT(stats.proj_3, xfregs.rawProjection[3]);
-			SETSTAT_FT(stats.proj_4, xfregs.rawProjection[4]);
-			SETSTAT_FT(stats.proj_5, xfregs.rawProjection[5]);
-			SETSTAT_FT(stats.proj_6, xfregs.rawProjection[6]);
+			SETSTAT_FT(stats.proj_0, rawProjection[0]);
+			SETSTAT_FT(stats.proj_1, rawProjection[1]);
+			SETSTAT_FT(stats.proj_2, rawProjection[2]);
+			SETSTAT_FT(stats.proj_3, rawProjection[3]);
+			SETSTAT_FT(stats.proj_4, rawProjection[4]);
+			SETSTAT_FT(stats.proj_5, rawProjection[5]);
+			break;
+			
+		default:
+			ERROR_LOG(VIDEO, "unknown projection type: %d", xfregs.projection.type);
 		}
 
-		PRIM_LOG("Projection: %f %f %f %f %f %f\n", xfregs.rawProjection[0], xfregs.rawProjection[1], xfregs.rawProjection[2], xfregs.rawProjection[3], xfregs.rawProjection[4], xfregs.rawProjection[5]);
+		PRIM_LOG("Projection: %f %f %f %f %f %f\n", rawProjection[0], rawProjection[1], rawProjection[2], rawProjection[3], rawProjection[4], rawProjection[5]);
 
-		if ((g_ActiveConfig.bFreeLook || g_ActiveConfig.bAnaglyphStereo ) && xfregs.rawProjection[6] == 0)
+		if ((g_ActiveConfig.bFreeLook || g_ActiveConfig.bAnaglyphStereo ) && xfregs.projection.type == GX_PERSPECTIVE)
 		{
 			Matrix44 mtxA;
 			Matrix44 mtxB;
diff --git a/Source/Core/VideoCommon/Src/VideoConfig.cpp b/Source/Core/VideoCommon/Src/VideoConfig.cpp
index 0e43904ab0..f17c33f2a7 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.cpp
+++ b/Source/Core/VideoCommon/Src/VideoConfig.cpp
@@ -75,7 +75,6 @@ void VideoConfig::Load(const char *ini_file)
 	iniFile.Get("Settings", "AnaglyphStereoSeparation", &iAnaglyphStereoSeparation, 200);
 	iniFile.Get("Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle, 0);
 	iniFile.Get("Settings", "EnablePixelLighting", &bEnablePixelLighting, 0);
-	iniFile.Get("Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth, 0);
 	
 	iniFile.Get("Settings", "MSAA", &iMultisampleMode, 0);
 	iniFile.Get("Settings", "EFBScale", &iEFBScale, 2); // native
@@ -134,7 +133,6 @@ void VideoConfig::GameIniLoad(const char *ini_file)
 	iniFile.GetIfExists("Video_Settings", "AnaglyphStereoSeparation", &iAnaglyphStereoSeparation);
 	iniFile.GetIfExists("Video_Settings", "AnaglyphFocalAngle", &iAnaglyphFocalAngle);
 	iniFile.GetIfExists("Video_Settings", "EnablePixelLighting", &bEnablePixelLighting);
-	iniFile.GetIfExists("Video_Settings", "EnablePerPixelDepth", &bEnablePerPixelDepth);
 	iniFile.GetIfExists("Video_Settings", "MSAA", &iMultisampleMode);
 	iniFile.GetIfExists("Video_Settings", "EFBScale", &iEFBScale); // integral
 	iniFile.GetIfExists("Video_Settings", "DstAlphaPass", &bDstAlphaPass);
@@ -204,7 +202,6 @@ void VideoConfig::Save(const char *ini_file)
 	iniFile.Set("Settings", "AnaglyphStereoSeparation", iAnaglyphStereoSeparation);
 	iniFile.Set("Settings", "AnaglyphFocalAngle", iAnaglyphFocalAngle);
 	iniFile.Set("Settings", "EnablePixelLighting", bEnablePixelLighting);
-	iniFile.Set("Settings", "EnablePerPixelDepth", bEnablePerPixelDepth);
 	
 
 	iniFile.Set("Settings", "ShowEFBCopyRegions", bShowEFBCopyRegions);
@@ -270,7 +267,6 @@ 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", "EnablePerPixelDepth", bEnablePerPixelDepth);
 	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 7b141d442a..d09f6995d6 100644
--- a/Source/Core/VideoCommon/Src/VideoConfig.h
+++ b/Source/Core/VideoCommon/Src/VideoConfig.h
@@ -133,7 +133,6 @@ struct VideoConfig
 	bool bZTPSpeedHack; // The Legend of Zelda: Twilight Princess
 	bool bUseBBox;
 	bool bEnablePixelLighting;
-	bool bEnablePerPixelDepth;
 
 	int iLog; // CONF_ bits
 	int iSaveTargetId; // TODO: Should be dropped
diff --git a/Source/Core/VideoCommon/Src/XFMemory.h b/Source/Core/VideoCommon/Src/XFMemory.h
index ee3c3b0fdc..fc199b8c32 100644
--- a/Source/Core/VideoCommon/Src/XFMemory.h
+++ b/Source/Core/VideoCommon/Src/XFMemory.h
@@ -61,6 +61,9 @@
 #define LIGHTATTN_NONE 2
 #define LIGHTATTN_DIR  3
 
+#define GX_PERSPECTIVE  0
+#define GX_ORTHOGRAPHIC 1
+
 #define XFMEM_SIZE               0x8000
 #define XFMEM_POSMATRICES        0x000
 #define XFMEM_POSMATRICES_END    0x100
@@ -235,6 +238,12 @@ struct Viewport
 	float farZ;
 };
 
+struct Projection
+{
+	float rawProjection[6];
+	u32 type;                      // only GX_PERSPECTIVE or GX_ORTHOGRAPHIC are allowed
+};
+
 struct XFRegisters
 {
     u32 error;                      // 0x1000
@@ -257,10 +266,10 @@ struct XFRegisters
     u32 unk5;                       // 0x1015
     u32 unk6;                       // 0x1016
     u32 unk7;                       // 0x1017
-    u32 MatrixIndexA;				// 0x1018
-    u32 MatrixIndexB;				// 0x1019
+    u32 MatrixIndexA;               // 0x1018
+    u32 MatrixIndexB;               // 0x1019
     Viewport viewport;              // 0x101a - 0x101f
-    float rawProjection[7];         // 0x1020 - 0x1026
+    Projection projection;          // 0x1020 - 0x1026
     u32 unk8[24];                   // 0x1027 - 0x103e
     NumTexGen numTexGen;            // 0x103f
     TexMtxInfo texMtxInfo[8];       // 0x1040 - 0x1047