Merge branch 'master' into vertex-loader-cleanup

Conflicts:
	Source/Core/Common/Src/CommonFuncs.h
	Source/Core/VideoCommon/Src/VertexLoader.cpp

Source/Core/VideoCommon/CMakeLists.txt

@@ -3,7 +3,6 @@ set(SRCS	Src/BPFunctions.cpp
 			Src/BPStructs.cpp
 			Src/CPMemory.cpp
 			Src/CommandProcessor.cpp
-			Src/DLCache.cpp
 			Src/Debugger.cpp
 			Src/Fifo.cpp
 			Src/FPSCounter.cpp
@@ -17,6 +16,7 @@ set(SRCS	Src/BPFunctions.cpp
 			Src/OpcodeDecoding.cpp
 			Src/OpenCL.cpp
 			Src/OpenCL/OCLTextureDecoder.cpp
+			Src/PerfQueryBase.cpp
 			Src/PixelEngine.cpp
 			Src/PixelShaderGen.cpp
 			Src/PixelShaderManager.cpp
@@ -24,7 +24,6 @@ set(SRCS	Src/BPFunctions.cpp
 			Src/Statistics.cpp
 			Src/TextureCacheBase.cpp
 			Src/TextureConversionShader.cpp
-			Src/TextureDecoder.cpp
 			Src/VertexLoader.cpp
 			Src/VertexLoaderManager.cpp
 			Src/VertexLoader_Color.cpp
@@ -41,6 +40,14 @@ set(SRCS	Src/BPFunctions.cpp
 			Src/memcpy_amd.cpp)
 
 set(LIBS core)
+
+if(NOT _M_GENERIC)
+	set(SRCS ${SRCS}	Src/x64TextureDecoder.cpp
+						Src/x64DLCache.cpp)
+else()
+	set(SRCS ${SRCS}	Src/GenericTextureDecoder.cpp
+						Src/GenericDLCache.cpp)
+endif()
 if(NOT ${CL} STREQUAL CL-NOTFOUND)
 	list(APPEND LIBS ${CL})
 endif()

Source/Core/VideoCommon/Src/AVIDump.cpp

@@ -157,9 +157,21 @@ void AVIDump::Stop()
 	NOTICE_LOG(VIDEO, "Stop");
 }
 
-void AVIDump::AddFrame(char *data)
+void AVIDump::AddFrame(const u8* data, int w, int h)
 {
-	AVIStreamWrite(m_streamCompressed, ++m_frameCount, 1, (LPVOID) data, m_bitmap.biSizeImage, AVIIF_KEYFRAME, NULL, &m_byteBuffer);
+	static bool shown_error = false;
+	if ((w != m_bitmap.biWidth || h != m_bitmap.biHeight) && !shown_error)
+	{
+		PanicAlert("You have resized the window while dumping frames.\n"
+			"Nothing sane can be done to handle this.\n"
+			"Your video will likely be broken.");
+		shown_error = true;
+
+		m_bitmap.biWidth = w;
+		m_bitmap.biHeight = h;
+	}
+
+	AVIStreamWrite(m_streamCompressed, ++m_frameCount, 1, const_cast<u8*>(data), m_bitmap.biSizeImage, AVIIF_KEYFRAME, NULL, &m_byteBuffer);
 	m_totalBytes += m_byteBuffer;
 	// Close the recording if the file is more than 2gb
 	// VfW can't properly save files over 2gb in size, but can keep writing to them up to 4gb.
@@ -298,9 +310,9 @@ bool AVIDump::CreateFile()
 	return true;
 }
 
-void AVIDump::AddFrame(uint8_t *data, int width, int height)
+void AVIDump::AddFrame(const u8* data, int width, int height)
 {
-	avpicture_fill((AVPicture *)s_BGRFrame, data, PIX_FMT_BGR24, width, height);
+	avpicture_fill((AVPicture *)s_BGRFrame, const_cast<u8*>(data), PIX_FMT_BGR24, width, height);
 
 	// Convert image from BGR24 to desired pixel format, and scale to initial
 	// width and height

Source/Core/VideoCommon/Src/AVIDump.h

@@ -24,6 +24,8 @@
 #include <stdint.h>
 #endif
 
+#include "CommonTypes.h"
+
 class AVIDump
 {
 	private:
@@ -36,11 +38,11 @@ class AVIDump
 	public:
 #ifdef _WIN32
 		static bool Start(HWND hWnd, int w, int h);
-		static void AddFrame(char *data);
 #else
 		static bool Start(int w, int h);
-		static void AddFrame(uint8_t *data, int width, int height);
 #endif
+		static void AddFrame(const u8* data, int width, int height);
+		
 		static void Stop();
 };
 

Source/Core/VideoCommon/Src/BPMemory.h

@@ -62,7 +62,7 @@
 #define BPMEM_COPYFILTER1      0x54
 #define BPMEM_CLEARBBOX1       0x55 
 #define BPMEM_CLEARBBOX2       0x56
-#define BPMEM_UNKNOWN_57       0x57
+#define BPMEM_CLEAR_PIXEL_PERF 0x57
 #define BPMEM_REVBITS          0x58
 #define BPMEM_SCISSOROFFSET    0x59
 #define BPMEM_PRELOAD_ADDR     0x60

Source/Core/VideoCommon/Src/BPStructs.cpp

@@ -31,6 +31,7 @@
 #include "VertexShaderManager.h"
 #include "Thread.h"
 #include "HW/Memmap.h"
+#include "PerfQueryBase.h"
 
 using namespace BPFunctions;
 
@@ -62,7 +63,6 @@ void RenderToXFB(const BPCmd &bp, const EFBRectangle &rc, float yScale, float xf
 {
 	Renderer::RenderToXFB(xfbAddr, dstWidth, dstHeight, rc, gamma);
 }
-
 void BPWritten(const BPCmd& bp)
 {
 	/*
@@ -144,7 +144,8 @@ void BPWritten(const BPCmd& bp)
 					|| bp.address == BPMEM_LOADTLUT0
 					|| bp.address == BPMEM_LOADTLUT1
 					|| bp.address == BPMEM_TEXINVALIDATE
-					|| bp.address == BPMEM_PRELOAD_MODE))
+					|| bp.address == BPMEM_PRELOAD_MODE
+					|| bp.address == BPMEM_CLEAR_PIXEL_PERF))
 			{
 				return;
 			}
@@ -484,9 +485,10 @@ void BPWritten(const BPCmd& bp)
 	case BPMEM_IND_IMASK: // Index Mask ?
 	case BPMEM_REVBITS: // Always set to 0x0F when GX_InitRevBits() is called.
 		break;
-
-	case BPMEM_UNKNOWN_57: // Sunshine alternates this register between values 0x000 and 0xAAA
-		DEBUG_LOG(VIDEO, "Unknown BP Reg 0x57: %08x", bp.newvalue);
+ 
+	case BPMEM_CLEAR_PIXEL_PERF:
+		// GXClearPixMetric writes 0xAAA here, Sunshine alternates this register between values 0x000 and 0xAAA
+		g_perf_query->ResetQuery();
 		break;
 
 	case BPMEM_PRELOAD_ADDR:

Source/Core/VideoCommon/Src/Fifo.cpp

@@ -222,11 +222,11 @@ void RunGpu()
 		{
 			u8 *uData = Memory::GetPointer(fifo.CPReadPointer);			
 			
-			SaveSSEState();
-			LoadDefaultSSEState();
+			FPURoundMode::SaveSIMDState();
+			FPURoundMode::LoadDefaultSIMDState();
 			ReadDataFromFifo(uData, 32);				
 			OpcodeDecoder_Run(g_bSkipCurrentFrame);	
-			LoadSSEState();
+			FPURoundMode::LoadSIMDState();
 
 			//DEBUG_LOG(COMMANDPROCESSOR, "Fifo wraps to base");
 

Source/Core/VideoCommon/Src/GenericDLCache.cpp

@@ -0,0 +1,52 @@
+// Copyright (C) 2003-2009 Dolphin Project.
+
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, version 2.0.
+
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License 2.0 for more details.
+
+// A copy of the GPL 2.0 should have been included with the program.
+// If not, see http://www.gnu.org/licenses/
+
+// Official SVN repository and contact information can be found at
+// http://code.google.com/p/dolphin-emu/
+
+// TODO: Handle cache-is-full condition :p
+
+
+#include "Common.h"
+#include "DLCache.h"
+
+namespace DLCache
+{
+	
+void Init()
+{
+}
+
+void Shutdown()
+{
+}
+
+void Clear() 
+{	
+}
+
+void ProgressiveCleanup()
+{
+}
+}  // namespace
+
+// NOTE - outside the namespace on purpose.
+bool HandleDisplayList(u32 address, u32 size)
+{
+	return false;
+}
+
+void IncrementCheckContextId()
+{
+}

Source/Core/VideoCommon/Src/ImageWrite.cpp

@@ -69,7 +69,8 @@ bool SaveTGA(const char* filename, int width, int height, void* pdata)
 
 bool SaveData(const char* filename, const char* data)
 {
-	std::ofstream f(filename, std::ios::binary);
+	std::ofstream f;
+	OpenFStream(f, filename, std::ios::binary);
 	f << data;
 
 	return true;

Source/Core/VideoCommon/Src/MainBase.cpp

@@ -21,6 +21,10 @@ volatile u32 s_swapRequested = false;
 u32 s_efbAccessRequested = false;
 volatile u32 s_FifoShuttingDown = false;
 
+std::condition_variable s_perf_query_cond;
+std::mutex s_perf_query_lock;
+static volatile bool s_perf_query_requested;
+
 static volatile struct
 {
 	u32 xfbAddr;
@@ -169,6 +173,43 @@ u32 VideoBackendHardware::Video_AccessEFB(EFBAccessType type, u32 x, u32 y, u32
 	return 0;
 }
 
+static bool QueryResultIsReady()
+{
+	return !s_perf_query_requested || s_FifoShuttingDown;
+}
+
+void VideoFifo_CheckPerfQueryRequest()
+{
+	if (s_perf_query_requested)
+	{
+		g_perf_query->FlushResults();
+		
+		{
+		std::lock_guard<std::mutex> lk(s_perf_query_lock);
+		s_perf_query_requested = false;
+		}
+		
+		s_perf_query_cond.notify_one();
+	}
+}
+
+u32 VideoBackendHardware::Video_GetQueryResult(PerfQueryType type)
+{
+	// TODO: Is this check sane?
+	if (!g_perf_query->IsFlushed())
+	{
+		if (SConfig::GetInstance().m_LocalCoreStartupParameter.bCPUThread)
+		{
+			s_perf_query_requested = true;
+			std::unique_lock<std::mutex> lk(s_perf_query_lock);
+			s_perf_query_cond.wait(lk, QueryResultIsReady);
+		}
+		else
+			g_perf_query->FlushResults();
+	}
+
+	return g_perf_query->GetQueryResult(type);
+}
 
 void VideoBackendHardware::InitializeShared()
 {
@@ -176,6 +217,7 @@ void VideoBackendHardware::InitializeShared()
 
 	s_swapRequested = 0;
 	s_efbAccessRequested = 0;
+	s_perf_query_requested = false;
 	s_FifoShuttingDown = 0;
 	memset((void*)&s_beginFieldArgs, 0, sizeof(s_beginFieldArgs));
 	memset(&s_accessEFBArgs, 0, sizeof(s_accessEFBArgs));
@@ -186,6 +228,11 @@ void VideoBackendHardware::InitializeShared()
 // Run from the CPU thread
 void VideoBackendHardware::DoState(PointerWrap& p)
 {
+	bool software = false;
+	p.Do(software);
+	if (p.GetMode() == PointerWrap::MODE_READ && software == true)
+		// change mode to abort load of incompatible save state.
+		p.SetMode(PointerWrap::MODE_VERIFY);
 	VideoCommon_DoState(p);
 	p.DoMarker("VideoCommon");
 
@@ -233,6 +280,7 @@ void VideoFifo_CheckAsyncRequest()
 {
 	VideoFifo_CheckSwapRequest();
 	VideoFifo_CheckEFBAccess();
+	VideoFifo_CheckPerfQueryRequest();
 }
 
 void VideoBackendHardware::Video_GatherPipeBursted()

Source/Core/VideoCommon/Src/OnScreenDisplay.h

@@ -22,7 +22,7 @@ namespace OSD
 {
 
 // On-screen message display
-void AddMessage(const char* str, u32 ms);
+void AddMessage(const char* str, u32 ms = 2000);
 void DrawMessages(); // draw the current messages on the screen. Only call once per frame.
 void ClearMessages();
 

Source/Core/VideoCommon/Src/PerfQueryBase.cpp

@@ -0,0 +1,3 @@
+#include "PerfQueryBase.h"
+
+PerfQueryBase* g_perf_query = 0;

Source/Core/VideoCommon/Src/PerfQueryBase.h

@@ -0,0 +1,54 @@
+#ifndef _PERFQUERY_BASE_H_
+#define _PERFQUERY_BASE_H_
+
+#include "CommonTypes.h"
+
+enum PerfQueryType
+{
+	PQ_ZCOMP_INPUT_ZCOMPLOC = 0,
+	PQ_ZCOMP_OUTPUT_ZCOMPLOC,
+	PQ_ZCOMP_INPUT,
+	PQ_ZCOMP_OUTPUT,
+	PQ_BLEND_INPUT,
+	PQ_EFB_COPY_CLOCKS,
+	PQ_NUM_MEMBERS
+};
+
+enum PerfQueryGroup
+{
+	PQG_ZCOMP_ZCOMPLOC,
+	PQG_ZCOMP,
+	PQG_EFB_COPY_CLOCKS,
+	PQG_NUM_MEMBERS,
+};
+
+class PerfQueryBase
+{
+public:
+	PerfQueryBase() {};
+	virtual ~PerfQueryBase() {}
+
+	// Begin querying the specified value for the following host GPU commands
+	virtual void EnableQuery(PerfQueryGroup type) {}
+
+	// Stop querying the specified value for the following host GPU commands
+	virtual void DisableQuery(PerfQueryGroup type) {}
+
+	// Reset query counters to zero and drop any pending queries
+	virtual void ResetQuery() {}
+
+	// Return the measured value for the specified query type
+	// NOTE: Called from CPU thread
+	virtual u32 GetQueryResult(PerfQueryType type) { return 0; }
+
+	// Request the value of any pending queries - causes a pipeline flush and thus should be used carefully!
+	virtual void FlushResults() {}
+
+	// True if there are no further pending query results
+	// NOTE: Called from CPU thread
+	virtual bool IsFlushed() const { return true; }
+};
+
+extern PerfQueryBase* g_perf_query;
+
+#endif // _PERFQUERY_H_

Source/Core/VideoCommon/Src/PixelEngine.cpp

@@ -28,10 +28,13 @@
 #include "ConfigManager.h"
 
 #include "PixelEngine.h"
+#include "RenderBase.h"
 #include "CommandProcessor.h"
 #include "HW/ProcessorInterface.h"
 #include "DLCache.h"
 #include "State.h"
+#include "PerfQueryBase.h"
+
 namespace PixelEngine
 {
 
@@ -255,23 +258,59 @@ void Read16(u16& _uReturnValue, const u32 _iAddress)
 		break;
 	}
 
-	case PE_PERF_0L:
-	case PE_PERF_0H:
-	case PE_PERF_1L:
-	case PE_PERF_1H:
-	case PE_PERF_2L:
-	case PE_PERF_2H:
-	case PE_PERF_3L:
-	case PE_PERF_3H:
-	case PE_PERF_4L:
-	case PE_PERF_4H:
-	case PE_PERF_5L:
-	case PE_PERF_5H:
-		INFO_LOG(PIXELENGINE, "(r16) perf counter @ %08x", _iAddress);
-		// git r90a2096a24f4 (svn r3663) added the PE_PERF cases, without setting
-		// _uReturnValue to anything, this reverts to the previous behaviour which allows
-		// The timer in SMS:Scrubbing Serena Beach to countdown correctly
-		_uReturnValue = 1;
+	// NOTE(neobrain): only PE_PERF_ZCOMP_OUTPUT is implemented in D3D11, but the other values shouldn't be contradictionary to the value of that register (i.e. INPUT registers should always be greater or equal to their corresponding OUTPUT registers).
+	case PE_PERF_ZCOMP_INPUT_ZCOMPLOC_L:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_INPUT_ZCOMPLOC) & 0xFFFF;
+		break;
+
+	case PE_PERF_ZCOMP_INPUT_ZCOMPLOC_H:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_INPUT_ZCOMPLOC) >> 16;
+		break;
+
+	case PE_PERF_ZCOMP_OUTPUT_ZCOMPLOC_L:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_OUTPUT_ZCOMPLOC) & 0xFFFF;
+		break;
+
+	case PE_PERF_ZCOMP_OUTPUT_ZCOMPLOC_H:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_OUTPUT_ZCOMPLOC) >> 16;
+		break;
+
+	case PE_PERF_ZCOMP_INPUT_L:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_INPUT) & 0xFFFF;
+		break;
+
+	case PE_PERF_ZCOMP_INPUT_H:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_INPUT) >> 16;
+		break;
+
+	case PE_PERF_ZCOMP_OUTPUT_L:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_OUTPUT) & 0xFFFF;
+		break;
+
+	case PE_PERF_ZCOMP_OUTPUT_H:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_ZCOMP_OUTPUT) >> 16;
+		break;
+
+	case PE_PERF_BLEND_INPUT_L:
+		// Super Mario Sunshine uses this register in episode 6 of Sirena Beach:
+		// The amount of remaining goop is determined by checking how many pixels reach the blending stage.
+		// Once this register falls below a particular value (around 0x90), the game regards the challenge finished.
+		// In very old builds, Dolphin only returned 0. That caused the challenge to be immediately finished without any goop being cleaned (the timer just didn't even start counting from 3:00:00).
+		// Later builds returned 1 for the high register. That caused the timer to actually count down, but made the challenge unbeatable because the game always thought you didn't clear any goop at all.
+		// Note that currently this functionality is only implemented in the D3D11 backend.
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_BLEND_INPUT) & 0xFFFF;
+		break;
+
+	case PE_PERF_BLEND_INPUT_H:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_BLEND_INPUT) >> 16;
+		break;
+
+	case PE_PERF_EFB_COPY_CLOCKS_L:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_EFB_COPY_CLOCKS) & 0xFFFF;
+		break;
+
+	case PE_PERF_EFB_COPY_CLOCKS_H:
+		_uReturnValue = g_video_backend->Video_GetQueryResult(PQ_EFB_COPY_CLOCKS) >> 16;
 		break;
 
 	default:

Source/Core/VideoCommon/Src/PixelEngine.h

@@ -36,19 +36,20 @@ enum
 	PE_BBOX_TOP		 = 0x14, // Flip Top
 	PE_BBOX_BOTTOM	 = 0x16, // Flip Bottom
 
-	// These have not yet been RE:d. They are the perf counters.
-	PE_PERF_0L       = 0x18, 
-	PE_PERF_0H       = 0x1a, 
-	PE_PERF_1L       = 0x1c, 
-	PE_PERF_1H       = 0x1e, 
-	PE_PERF_2L       = 0x20, 
-	PE_PERF_2H       = 0x22, 
-	PE_PERF_3L       = 0x24, 
-	PE_PERF_3H       = 0x26, 
-	PE_PERF_4L       = 0x28, 
-	PE_PERF_4H       = 0x2a, 
-	PE_PERF_5L       = 0x2c, 
-	PE_PERF_5H       = 0x2e, 
+	// NOTE: Order not verified
+	// These indicate the number of quads that are being used as input/output for each particular stage
+	PE_PERF_ZCOMP_INPUT_ZCOMPLOC_L	= 0x18, 
+	PE_PERF_ZCOMP_INPUT_ZCOMPLOC_H	= 0x1a,
+	PE_PERF_ZCOMP_OUTPUT_ZCOMPLOC_L	= 0x1c, 
+	PE_PERF_ZCOMP_OUTPUT_ZCOMPLOC_H	= 0x1e, 
+	PE_PERF_ZCOMP_INPUT_L			= 0x20, 
+	PE_PERF_ZCOMP_INPUT_H			= 0x22,
+	PE_PERF_ZCOMP_OUTPUT_L			= 0x24, 
+	PE_PERF_ZCOMP_OUTPUT_H			= 0x26, 
+	PE_PERF_BLEND_INPUT_L			= 0x28, 
+	PE_PERF_BLEND_INPUT_H			= 0x2a, 
+	PE_PERF_EFB_COPY_CLOCKS_L		= 0x2c, 
+	PE_PERF_EFB_COPY_CLOCKS_H		= 0x2e, 
 };
 
 namespace PixelEngine

Source/Core/VideoCommon/Src/PixelShaderGen.cpp

@@ -252,7 +252,8 @@ void ValidatePixelShaderIDs(API_TYPE api, PIXELSHADERUIDSAFE old_id, const std::
 			static int num_failures = 0;
 			char szTemp[MAX_PATH];
 			sprintf(szTemp, "%spsuid_mismatch_%04i.txt", File::GetUserPath(D_DUMP_IDX).c_str(), num_failures++);
-			std::ofstream file(szTemp);
+			std::ofstream file;
+			OpenFStream(file, szTemp, std::ios_base::out);
 			file << msg;
 			file << "\n\nOld shader code:\n" << old_code;
 			file << "\n\nNew shader code:\n" << new_code;

Source/Core/VideoCommon/Src/RenderBase.cpp

@@ -83,7 +83,9 @@ unsigned int Renderer::efb_scale_denominatorY = 1;
 unsigned int Renderer::ssaa_multiplier = 1;
 
 
-Renderer::Renderer() : frame_data(NULL), bLastFrameDumped(false)
+Renderer::Renderer()
+	: frame_data()
+	, bLastFrameDumped(false)
 {
 	UpdateActiveConfig();
 	TextureCache::OnConfigChanged(g_ActiveConfig);
@@ -110,7 +112,6 @@ Renderer::~Renderer()
 	if (pFrameDump.IsOpen())
 		pFrameDump.Close();
 #endif
-	delete[] frame_data;
 }
 
 void Renderer::RenderToXFB(u32 xfbAddr, u32 fbWidth, u32 fbHeight, const EFBRectangle& sourceRc, float Gamma)

Source/Core/VideoCommon/Src/RenderBase.h

@@ -52,6 +52,15 @@ public:
 	Renderer();
 	virtual ~Renderer();
 
+	enum PixelPerfQuery {
+		PP_ZCOMP_INPUT_ZCOMPLOC,
+		PP_ZCOMP_OUTPUT_ZCOMPLOC,
+		PP_ZCOMP_INPUT,
+		PP_ZCOMP_OUTPUT,
+		PP_BLEND_INPUT,
+		PP_EFB_COPY_CLOCKS
+	};
+
 	virtual void SetColorMask() = 0;
 	virtual void SetBlendMode(bool forceUpdate) = 0;
 	virtual void SetScissorRect(const TargetRectangle& rc) = 0;
@@ -147,7 +156,7 @@ protected:
 #else
 	File::IOFile pFrameDump;
 #endif
-	char* frame_data;
+	std::vector<u8> frame_data;
 	bool bLastFrameDumped;
 
 	// The framebuffer size

Source/Core/VideoCommon/Src/TextureCacheBase.cpp

@@ -132,8 +132,7 @@ void TextureCache::Cleanup()
 		if (	frameCount > TEXTURE_KILL_THRESHOLD + iter->second->frameCount
 			
 			// EFB copies living on the host GPU are unrecoverable and thus shouldn't be deleted
-			// TODO: encoding the texture back to RAM here might be a good idea
-			&& ! (g_ActiveConfig.bCopyEFBToTexture && iter->second->IsEfbCopy()) )
+			&& ! iter->second->IsEfbCopy() )
 		{
 			delete iter->second;
 			textures.erase(iter++);

Source/Core/VideoCommon/Src/VertexLoader.cpp

@@ -23,7 +23,7 @@
 #include "MemoryUtil.h"
 #include "StringUtil.h"
 #include "x64Emitter.h"
-#include "ABI.h"
+#include "x64ABI.h"
 #include "PixelEngine.h"
 #include "Host.h"
 
@@ -43,8 +43,9 @@
 //BBox
 #include "XFMemory.h"
 extern float GC_ALIGNED16(g_fProjectionMatrix[16]);
-
+#ifndef _M_GENERIC
 #define USE_JIT
+#endif
 
 #define COMPILED_CODE_SIZE 4096
 
@@ -86,8 +87,9 @@ static const float fractionTable[32] = {
 	1.0f / (1U << 24), 1.0f / (1U << 25), 1.0f / (1U << 26), 1.0f / (1U << 27),
 	1.0f / (1U << 28), 1.0f / (1U << 29), 1.0f / (1U << 30), 1.0f / (1U << 31),
 };
-
+#ifdef USE_JIT
 using namespace Gen;
+#endif
 
 void LOADERDECL PosMtx_ReadDirect_UByte()
 {
@@ -199,14 +201,19 @@ VertexLoader::VertexLoader(const TVtxDesc &vtx_desc, const VAT &vtx_attr)
 	m_VtxDesc = vtx_desc;
 	SetVAT(vtx_attr.g0.Hex, vtx_attr.g1.Hex, vtx_attr.g2.Hex);
 
+	#ifdef USE_JIT
 	AllocCodeSpace(COMPILED_CODE_SIZE);
 	CompileVertexTranslator();
 	WriteProtect();
+	#endif
+
 }
 
 VertexLoader::~VertexLoader() 
 {
+	#ifdef USE_JIT
 	FreeCodeSpace();
+	#endif
 	delete m_NativeFmt;
 }
 
@@ -492,7 +499,8 @@ void VertexLoader::WriteCall(TPipelineFunction func)
 	m_PipelineStages[m_numPipelineStages++] = func;
 #endif
 }
-
+// ARMTODO: This should be done in a better way
+#ifndef _M_GENERIC
 void VertexLoader::WriteGetVariable(int bits, OpArg dest, void *address)
 {
 #ifdef USE_JIT
@@ -516,6 +524,7 @@ void VertexLoader::WriteSetVariable(int bits, void *address, OpArg value)
 #endif
 #endif
 }
+#endif
 
 int VertexLoader::SetupRunVertices(int vtx_attr_group, int primitive, int const count)
 {

Source/Core/VideoCommon/Src/VertexLoader.h

@@ -76,7 +76,12 @@ private:
 	}
 };
 
+// ARMTODO: This should be done in a better way
+#ifndef _M_GENERIC
 class VertexLoader : public Gen::XCodeBlock, NonCopyable
+#else
+class VertexLoader
+#endif
 {
 public:
 	VertexLoader(const TVtxDesc &vtx_desc, const VAT &vtx_attr);
@@ -125,8 +130,10 @@ private:
 
 	void WriteCall(TPipelineFunction);
 
+#ifndef _M_GENERIC
 	void WriteGetVariable(int bits, Gen::OpArg dest, void *address);
 	void WriteSetVariable(int bits, void *address, Gen::OpArg dest);
+#endif
 };									  
 
 #endif

Source/Core/VideoCommon/Src/VertexManagerBase.cpp

@@ -203,7 +203,9 @@ void VertexManager::Flush()
 	//if (g_nativeVertexFmt)
 		g_nativeVertexFmt->SetupVertexPointers();
 
+	g_renderer->ResumePixelPerf(false);
 	g_vertex_manager->Draw(stride, false);
+	g_renderer->PausePixelPerf(false);
 
 	// run through vertex groups again to set alpha
 	if (false == g_ActiveConfig.bDstAlphaPass && bpmem.dstalpha.enable && bpmem.blendmode.alphaupdate)
@@ -223,10 +225,12 @@ void VertexManager::Flush()
 		// save the shaders
 		char strfile[255];
 		sprintf(strfile, "%sps%.3d.txt", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), g_ActiveConfig.iSaveTargetId);
-		std::ofstream fps(strfile);
+		std::ofstream fps;
+		OpenFStream(fps, strfile, std::ios_base::out);
 		fps << ps->strprog.c_str();
 		sprintf(strfile, "%svs%.3d.txt", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), g_ActiveConfig.iSaveTargetId);
-		std::ofstream fvs(strfile);
+		std::ofstream fvs;
+		OpenFStream(fvs, strfile, std::ios_base::out);
 		fvs << vs->strprog.c_str();
 	}
 

Source/Core/VideoCommon/Src/VertexShaderGen.cpp

@@ -114,7 +114,8 @@ void ValidateVertexShaderIDs(API_TYPE api, VERTEXSHADERUIDSAFE old_id, const std
 			static int num_failures = 0;
 			char szTemp[MAX_PATH];
 			sprintf(szTemp, "%svsuid_mismatch_%04i.txt", File::GetUserPath(D_DUMP_IDX).c_str(), num_failures++);
-			std::ofstream file(szTemp);
+			std::ofstream file;
+			OpenFStream(file, szTemp, std::ios_base::out);
 			file << msg;
 			file << "\n\nOld shader code:\n" << old_code;
 			file << "\n\nNew shader code:\n" << new_code;

Source/Core/VideoCommon/Src/VideoConfig.h

@@ -115,7 +115,7 @@ struct VideoConfig
 	int iAnaglyphStereoSeparation;
 	int iAnaglyphFocalAngle;
 	bool b3DVision;
-	
+
 	// Hacks
 	bool bEFBAccessEnable;
 	bool bDlistCachingEnable;

Source/Core/VideoCommon/Src/x64DLCache.cpp

@@ -35,7 +35,7 @@
 #include "VertexLoaderManager.h"
 #include "VertexManagerBase.h"
 #include "x64Emitter.h"
-#include "ABI.h"
+#include "x64ABI.h"
 
 #include "DLCache.h"
 #include "VideoConfig.h"

Source/Core/VideoCommon/Src/x64TextureDecoder.cpp

@@ -1119,20 +1119,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 							_mm_storeu_si128( (__m128i*)( dst+(y + iy+1) * width + x + 4 ), o4 );
 						}
 			}
-#if 0
-			// Reference C implementation:
-			for (int y = 0; y < height; y += 8)
-				for (int x = 0; x < width; x += 8)
-					for (int iy = 0; iy < 8; iy++, src += 4)
-						for (int ix = 0; ix < 4; ix++)
-						{
-							int val = src[ix];
-							u8 i1 = Convert4To8(val >> 4);
-							u8 i2 = Convert4To8(val & 0xF);
-							memset(dst+(y + iy) * width + x + ix * 2 , i1,4);
-							memset(dst+(y + iy) * width + x + ix * 2 + 1 , i2,4);
-						}
-#endif
 		}
 	   break;
 	case GX_TF_I8:  // speed critical
@@ -1248,26 +1234,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 						
 					}
 			}
-#if 0
-			// Reference C implementation
-			for (int y = 0; y < height; y += 4)
-				for (int x = 0; x < width; x += 8)
-					for (int iy = 0; iy < 4; ++iy, src += 8)
-					{
-						u32 *  newdst = dst + (y + iy)*width+x;
-						const u8 *  newsrc = src;
-						u8 srcval;
-
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = (newsrc++)[0]; (newdst++)[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-						srcval = newsrc[0]; newdst[0] = srcval | (srcval << 8) | (srcval << 16) | (srcval << 24);
-					}
-#endif
 		}
 		break;
 	case GX_TF_C8:
@@ -1380,20 +1346,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 							_mm_storeu_si128( (__m128i*)(dst + (y + iy) * width + x), r1 );
 						}
 			}
-#if 0
-			// Reference C implementation:
-			for (int y = 0; y < height; y += 4)
-				for (int x = 0; x < width; x += 4)
-					for (int iy = 0; iy < 4; iy++, src += 8)
-					{
-						u32 *ptr = dst + (y + iy) * width + x;
-						u16 *s = (u16 *)src;
-						ptr[0] = decodeIA8Swapped(s[0]);
-						ptr[1] = decodeIA8Swapped(s[1]);
-						ptr[2] = decodeIA8Swapped(s[2]);
-						ptr[3] = decodeIA8Swapped(s[3]);
-					}
-#endif
 		}
 		break;
 	case GX_TF_C14X2:
@@ -1493,18 +1445,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 						__m128i *ptr = (__m128i *)(dst + (y + iy) * width + x);
 						_mm_storeu_si128(ptr, abgr888x4);
 					}
-#if 0
-			// Reference C implementation.
-			for (int y = 0; y < height; y += 4)
-				for (int x = 0; x < width; x += 4)
-					for (int iy = 0; iy < 4; iy++, src += 8)
-					{
-						u32 *ptr = dst + (y + iy) * width + x;
-						u16 *s = (u16 *)src;
-						for(int j = 0; j < 4; j++)
-							*ptr++ = decode565RGBA(Common::swap16(*s++));
-					}
-#endif
 		}
 		break;
 	case GX_TF_RGB5A3:
@@ -1718,13 +1658,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 							}
 						}
 				}
-#if 0
-			// Reference C implementation:
-			for (int y = 0; y < height; y += 4)
-				for (int x = 0; x < width; x += 4)
-					for (int iy = 0; iy < 4; iy++, src += 8)
-						decodebytesRGB5A3rgba(dst+(y+iy)*width+x, (u16*)src);
-#endif
 		}
 		break;
 	case GX_TF_RGBA8:  // speed critical
@@ -1860,16 +1793,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 						_mm_storeu_si128(dst128, rgba11);
 					}				
 			}
-#if 0
-			// Reference C implementation.
-			for (int y = 0; y < height; y += 4)
-				for (int x = 0; x < width; x += 4)
-				{
-					for (int iy = 0; iy < 4; iy++)
-						decodebytesARGB8_4ToRgba(dst + (y+iy)*width + x, (u16*)src + 4 * iy, (u16*)src + 4 * iy + 16);
-					src += 64;
-				}
-#endif
 		}
 		break;
 	case GX_TF_CMPR:  // speed critical
@@ -2104,22 +2027,6 @@ PC_TexFormat TexDecoder_Decode_RGBA(u32 * dst, const u8 * src, int width, int he
 					}					
 				}
 			}
-#if 0
-			for (int y = 0; y < height; y += 8)
-			{
-				for (int x = 0; x < width; x += 8)
-				{
-					decodeDXTBlockRGBA((u32*)dst + y * width + x, (DXTBlock*)src, width);
-										src += sizeof(DXTBlock);
-					decodeDXTBlockRGBA((u32*)dst + y * width + x + 4, (DXTBlock*)src, width);
-										src += sizeof(DXTBlock);
-					decodeDXTBlockRGBA((u32*)dst + (y + 4) * width + x, (DXTBlock*)src, width);
-										src += sizeof(DXTBlock);
-					decodeDXTBlockRGBA((u32*)dst + (y + 4) * width + x + 4, (DXTBlock*)src, width);
-										src += sizeof(DXTBlock);
-				}
-			}
-#endif
 			break;
 		}
 	}

Source/Core/VideoCommon/VideoCommon.vcxproj

@@ -111,8 +111,8 @@
   <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
     <ClCompile>
       <AdditionalIncludeDirectories>..\Common\Src;..\Core\Src;..\..\..\Externals\SOIL;..\..\..\Externals\CLRun\include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
-	  <OpenMPSupport>false</OpenMPSupport>
-	</ClCompile>
+      <OpenMPSupport>false</OpenMPSupport>
+    </ClCompile>
     <Link>
       <GenerateDebugInformation>true</GenerateDebugInformation>
     </Link>
@@ -143,7 +143,7 @@
     <ClCompile>
       <AdditionalIncludeDirectories>..\Common\Src;..\Core\Src;..\..\..\Externals\SOIL;..\..\..\Externals\CLRun\include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
       <OpenMPSupport>false</OpenMPSupport>
-	</ClCompile>
+    </ClCompile>
     <Link>
       <GenerateDebugInformation>true</GenerateDebugInformation>
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
@@ -182,7 +182,6 @@
     <ClCompile Include="Src\CommandProcessor.cpp" />
     <ClCompile Include="Src\CPMemory.cpp" />
     <ClCompile Include="Src\Debugger.cpp" />
-    <ClCompile Include="Src\DLCache.cpp" />
     <ClCompile Include="Src\EmuWindow.cpp" />
     <ClCompile Include="Src\Fifo.cpp" />
     <ClCompile Include="Src\FPSCounter.cpp" />
@@ -197,6 +196,7 @@
     <ClCompile Include="Src\OpcodeDecoding.cpp" />
     <ClCompile Include="Src\OpenCL.cpp" />
     <ClCompile Include="Src\OpenCL\OCLTextureDecoder.cpp" />
+    <ClCompile Include="Src\PerfQueryBase.cpp" />
     <ClCompile Include="Src\PixelEngine.cpp" />
     <ClCompile Include="Src\PixelShaderGen.cpp" />
     <ClCompile Include="Src\PixelShaderManager.cpp" />
@@ -204,7 +204,6 @@
     <ClCompile Include="Src\Statistics.cpp" />
     <ClCompile Include="Src\TextureCacheBase.cpp" />
     <ClCompile Include="Src\TextureConversionShader.cpp" />
-    <ClCompile Include="Src\TextureDecoder.cpp" />
     <ClCompile Include="Src\VertexLoader.cpp" />
     <ClCompile Include="Src\VertexLoaderManager.cpp" />
     <ClCompile Include="Src\VertexLoader_Color.cpp" />
@@ -216,6 +215,8 @@
     <ClCompile Include="Src\VertexShaderManager.cpp" />
     <ClCompile Include="Src\VideoConfig.cpp" />
     <ClCompile Include="Src\VideoState.cpp" />
+    <ClCompile Include="Src\x64DLCache.cpp" />
+    <ClCompile Include="Src\x64TextureDecoder.cpp" />
     <ClCompile Include="Src\XFMemory.cpp" />
     <ClCompile Include="Src\XFStructs.cpp" />
   </ItemGroup>
@@ -244,6 +245,7 @@
     <ClInclude Include="Src\OpcodeDecoding.h" />
     <ClInclude Include="Src\OpenCL.h" />
     <ClInclude Include="Src\OpenCL\OCLTextureDecoder.h" />
+    <ClInclude Include="Src\PerfQueryBase.h" />
     <ClInclude Include="Src\PixelEngine.h" />
     <ClInclude Include="Src\PixelShaderGen.h" />
     <ClInclude Include="Src\PixelShaderManager.h" />

Source/Core/VideoCommon/VideoCommon.vcxproj.filters

@@ -5,9 +5,6 @@
     <ClCompile Include="Src\memcpy_amd.cpp" />
     <ClCompile Include="Src\PixelEngine.cpp" />
     <ClCompile Include="Src\VideoConfig.cpp" />
-    <ClCompile Include="Src\DLCache.cpp">
-      <Filter>Vertex Loading</Filter>
-    </ClCompile>
     <ClCompile Include="Src\VertexLoader.cpp">
       <Filter>Vertex Loading</Filter>
     </ClCompile>
@@ -92,9 +89,6 @@
     <ClCompile Include="Src\OpcodeDecoding.cpp">
       <Filter>Decoding</Filter>
     </ClCompile>
-    <ClCompile Include="Src\TextureDecoder.cpp">
-      <Filter>Decoding</Filter>
-    </ClCompile>
     <ClCompile Include="Src\Debugger.cpp">
       <Filter>Base</Filter>
     </ClCompile>
@@ -107,6 +101,9 @@
     <ClCompile Include="Src\MainBase.cpp">
       <Filter>Base</Filter>
     </ClCompile>
+    <ClCompile Include="Src\PerfQueryBase.cpp">
+      <Filter>Base</Filter>
+    </ClCompile>
     <ClCompile Include="Src\RenderBase.cpp">
       <Filter>Base</Filter>
     </ClCompile>
@@ -122,6 +119,12 @@
     <ClCompile Include="Src\FPSCounter.cpp">
       <Filter>Util</Filter>
     </ClCompile>
+    <ClCompile Include="Src\x64TextureDecoder.cpp">
+      <Filter>Decoding</Filter>
+    </ClCompile>
+    <ClCompile Include="Src\x64DLCache.cpp">
+      <Filter>Vertex Loading</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="Src\CommandProcessor.h" />
@@ -237,6 +240,9 @@
     <ClInclude Include="Src\MainBase.h">
       <Filter>Base</Filter>
     </ClInclude>
+    <ClInclude Include="Src\PerfQueryBase.h">
+      <Filter>Base</Filter>
+    </ClInclude>
     <ClInclude Include="Src\RenderBase.h">
       <Filter>Base</Filter>
     </ClInclude>
@@ -285,4 +291,4 @@
       <UniqueIdentifier>{e2a527a2-ccc8-4ab8-a93e-dd2628c0f3b6}</UniqueIdentifier>
     </Filter>
   </ItemGroup>
-</Project>
+</Project>