diff --git a/GPU/Directx9/FramebufferDX9.cpp b/GPU/Directx9/FramebufferDX9.cpp
index 8c6bb70c44..6f9814691b 100644
--- a/GPU/Directx9/FramebufferDX9.cpp
+++ b/GPU/Directx9/FramebufferDX9.cpp
@@ -129,7 +129,17 @@ namespace DX9 {
 		// by themselves.
 		ClearBuffer();
 		// TODO: Check / use D3DCAPS2_DYNAMICTEXTURES?
-		pD3Ddevice->CreateTexture(512, 272, 1, 0, D3DFMT(D3DFMT_A8R8G8B8), D3DPOOL_MANAGED, &drawPixelsTex_, NULL);
+		int usage = 0;
+		D3DPOOL pool = D3DPOOL_MANAGED;
+		if (pD3DdeviceEx) {
+			pool = D3DPOOL_DEFAULT;
+			usage = D3DUSAGE_DYNAMIC;
+		}
+		HRESULT hr = pD3Ddevice->CreateTexture(512, 272, 1, usage, D3DFMT(D3DFMT_A8R8G8B8), pool, &drawPixelsTex_, NULL);
+		if (!hr) {
+			drawPixelsTex_ = nullptr;
+			ERROR_LOG(G3D, "Failed to create drawpixels texture");
+		}
 		useBufferedRendering_ = g_Config.iRenderingMode != FB_NON_BUFFERED_MODE;
 	}
 
@@ -158,6 +168,10 @@ namespace DX9 {
 		u8 * convBuf = NULL;
 		D3DLOCKED_RECT rect;
 
+		if (!drawPixelsTex_) {
+			return;
+		}
+
 		drawPixelsTex_->LockRect(0, &rect, NULL, 0);
 
 		convBuf = (u8*)rect.pBits;
diff --git a/GPU/Directx9/TextureCacheDX9.cpp b/GPU/Directx9/TextureCacheDX9.cpp
index 6ca63c6ae8..a6e39c575d 100644
--- a/GPU/Directx9/TextureCacheDX9.cpp
+++ b/GPU/Directx9/TextureCacheDX9.cpp
@@ -1632,7 +1632,13 @@ void TextureCacheDX9::LoadTextureLevel(TexCacheEntry &entry, int level, bool rep
 			pD3Ddevice->SetTexture(0, entry.texture);
 		} else {
 			// Create texture
-			HRESULT hr = pD3Ddevice->CreateTexture(w, h, 1, 0, (D3DFORMAT)D3DFMT(dstFmt), D3DPOOL_MANAGED, &entry.texture, NULL);
+			D3DPOOL pool = D3DPOOL_MANAGED;
+			int usage = 0;
+			if (pD3DdeviceEx) {
+				pool = D3DPOOL_DEFAULT;
+				usage = D3DUSAGE_DYNAMIC;  // TODO: Switch to using a staging texture?
+			}
+			HRESULT hr = pD3Ddevice->CreateTexture(w, h, 1, usage, (D3DFORMAT)D3DFMT(dstFmt), pool, &entry.texture, NULL);
 			if (FAILED(hr)) {
 				INFO_LOG(G3D, "Failed to create D3D texture");
 			} else {
diff --git a/GPU/Directx9/VertexDecoderDX9.cpp b/GPU/Directx9/VertexDecoderDX9.cpp
index 44e3f04eb5..5cd7a9aa19 100644
--- a/GPU/Directx9/VertexDecoderDX9.cpp
+++ b/GPU/Directx9/VertexDecoderDX9.cpp
@@ -107,7 +107,7 @@ void VertexDecoderDX9::Step_WeightsU16() const
 		wt[j] *= (1.0f/65535.f);
 	}
 	while (j & 3)   // Zero additional weights rounding up to 4.
-		wt[j++] = 0;
+		wt[j++] = 0.0f;
 #else
 	u16 *wt = (u16 *)(decoded_  + decFmt.w0off);
 	const u16_le *wdata = (const u16_le*)(ptr_);
@@ -124,26 +124,14 @@ void VertexDecoderDX9::Step_WeightsU16() const
 // (PSP uses 0.0-2.0 fixed point numbers for weights)
 void VertexDecoderDX9::Step_WeightsFloat() const
 {
-#if 0
-	float *wt = (float *)(decoded_ + decFmt.w0off);
-	const float_le *wdata = (const float_le*)(ptr_);
-	int j;
-	for (j = 0; j < nweights; j++) {
-		wt[j] = wdata[j];
-	}
-	while (j & 3)   // Zero additional weights rounding up to 4.
-		wt[j++] = 0.0f;
-#else
-	float *wt = (float *)(decoded_ + decFmt.w0off);
-	u32 *st = (u32 *)wt;
+	u32 *st = (u32 *)(decoded_ + decFmt.w0off);
 	const u32_le *wdata = (const u32_le*)(ptr_);
 	int j;
 	for (j = 0; j < nweights; j++) {
 		st[j] = wdata[j];
 	}
 	while (j & 3)   // Zero additional weights rounding up to 4.
-		wt[j++] = 0.0f;
-#endif
+		st[j++] = 0;
 }
 
 void VertexDecoderDX9::Step_TcU8() const
diff --git a/GPU/Directx9/helper/global.cpp b/GPU/Directx9/helper/global.cpp
index e8fa5fa70e..97d42647d5 100644
--- a/GPU/Directx9/helper/global.cpp
+++ b/GPU/Directx9/helper/global.cpp
@@ -5,6 +5,7 @@
 namespace DX9 {
 
 LPDIRECT3DDEVICE9 pD3Ddevice = NULL;
+LPDIRECT3DDEVICE9EX pD3DdeviceEx = NULL;
 LPDIRECT3D9 pD3D = NULL;
 
 static const char * vscode =
diff --git a/GPU/Directx9/helper/global.h b/GPU/Directx9/helper/global.h
index abb52aa4bc..cc0a37f1b1 100644
--- a/GPU/Directx9/helper/global.h
+++ b/GPU/Directx9/helper/global.h
@@ -17,6 +17,7 @@ struct ID3DXConstantTable;
 namespace DX9 {
 
 extern LPDIRECT3DDEVICE9 pD3Ddevice;
+extern LPDIRECT3DDEVICE9EX pD3DdeviceEx;
 
 extern LPDIRECT3DVERTEXSHADER9      pFramebufferVertexShader; // Vertex Shader
 extern LPDIRECT3DPIXELSHADER9       pFramebufferPixelShader;  // Pixel Shader
diff --git a/UI/ControlMappingScreen.cpp b/UI/ControlMappingScreen.cpp
index 46955b31dc..4113e0dad1 100644
--- a/UI/ControlMappingScreen.cpp
+++ b/UI/ControlMappingScreen.cpp
@@ -360,34 +360,34 @@ public:
 			yAxis_(yAxis), yDevice_(yDevice), 
 			curX_(0.0f), curY_(0.0f),
 			maxCount_(500) {}
-	void Draw(UIContext &dc) override;
-	void Update(const InputState &input_state) override;
-	void Axis(const AxisInput &input) {
-		if (input.axisId == xAxis_) {
-			curX_ = input.value;
-		} else if (input.axisId == yAxis_) {
-			curY_ = input.value;
-		}
-	}
-
-private:
-	struct Location {
-		Location() : x(0.0f), y(0.0f) {}
-		Location(float xx, float yy) : x(xx), y(yy) {}
-		float x;
-		float y;
-	};
-
-	int xAxis_;
-	int yAxis_;
-	int xDevice_;
-	int yDevice_;
-
-	float curX_;
-	float curY_;
-
-	std::deque<Location> locations_;
-	int maxCount_;
+	void Draw(UIContext &dc) override;
+	void Update(const InputState &input_state) override;
+	void Axis(const AxisInput &input) {
+		if (input.axisId == xAxis_) {
+			curX_ = input.value;
+		} else if (input.axisId == yAxis_) {
+			curY_ = input.value;
+		}
+	}
+
+private:
+	struct Location {
+		Location() : x(0.0f), y(0.0f) {}
+		Location(float xx, float yy) : x(xx), y(yy) {}
+		float x;
+		float y;
+	};
+
+	int xAxis_;
+	int yAxis_;
+	int xDevice_;
+	int yDevice_;
+
+	float curX_;
+	float curY_;
+
+	std::deque<Location> locations_;
+	int maxCount_;
 };
 
 void JoystickHistoryView::Draw(UIContext &dc) {
diff --git a/Windows/D3D9Base.cpp b/Windows/D3D9Base.cpp
index 4823a1ebaf..04c62eb5ae 100644
--- a/Windows/D3D9Base.cpp
+++ b/Windows/D3D9Base.cpp
@@ -13,9 +13,12 @@
 #include "thin3d/thin3d.h"
 #include "thin3d/d3dx9_loader.h"
 
+static bool has9Ex = false;
 static LPDIRECT3D9 d3d;
+static LPDIRECT3D9EX d3dEx;
 static int adapterId;
 static LPDIRECT3DDEVICE9 device;
+static LPDIRECT3DDEVICE9EX deviceEx;
 static HDC hDC;     // Private GDI Device Context
 static HGLRC hRC;   // Permanent Rendering Context
 static HWND hWnd;   // Holds Our Window Handle
@@ -26,27 +29,57 @@ static int xres, yres;
 static bool enableGLDebug = true;
 
 void D3D9_SwapBuffers() {
-	device->EndScene();
-	device->Present(NULL, NULL, NULL, NULL);
-	device->BeginScene();
+	if (has9Ex) {
+		deviceEx->EndScene();
+		deviceEx->PresentEx(NULL, NULL, NULL, NULL, 0);
+		deviceEx->BeginScene();
+	} else {
+		device->EndScene();
+		device->Present(NULL, NULL, NULL, NULL);
+		device->BeginScene();
+	}
 }
 
 Thin3DContext *D3D9_CreateThin3DContext() {
-	return T3DCreateDX9Context(d3d, adapterId, device);
+	return T3DCreateDX9Context(d3d, d3dEx, adapterId, device, deviceEx);
+}
+
+typedef HRESULT (*DIRECT3DCREATE9EX)(UINT, IDirect3D9Ex**);
+
+bool IsWin7OrLater() {
+	DWORD version = GetVersion();
+	DWORD major = (DWORD)(LOBYTE(LOWORD(version)));
+	DWORD minor = (DWORD)(HIBYTE(LOWORD(version)));
+
+	return (major > 6) || ((major == 6) && (minor >= 1));
 }
 
 bool D3D9_Init(HWND hWnd, bool windowed, std::string *error_message) {
-	d3d = Direct3DCreate9(D3D_SDK_VERSION);
-	if (!d3d) {
-		ELOG("Failed to create D3D context");
-		return false;
+	DIRECT3DCREATE9EX g_pfnCreate9ex;
+
+	HMODULE hD3D9 = LoadLibrary(TEXT("d3d9.dll"));
+	g_pfnCreate9ex = (DIRECT3DCREATE9EX)GetProcAddress(hD3D9, "Direct3DCreate9Ex");
+	has9Ex = (g_pfnCreate9ex != NULL);
+	FreeLibrary(hD3D9);
+
+	// For some reason, can't quite get Ex to work right..
+	// has9Ex = false;
+
+	if (has9Ex) {
+		HRESULT result = Direct3DCreate9Ex(D3D_SDK_VERSION, &d3dEx);
+		d3d = d3dEx;
+		if (FAILED(result)) {
+			ELOG("Failed to create D3D9Ex context");
+			return false;
+		}
+	} else {
+		d3d = Direct3DCreate9(D3D_SDK_VERSION);
+		if (!d3d) {
+			ELOG("Failed to create D3D context");
+			return false;
+		}
 	}
 
-	RECT rc;
-	GetClientRect(hWnd, &rc);
-	int xres = rc.right - rc.left;
-	int yres = rc.bottom - rc.top;
-	
 	D3DCAPS9 d3dCaps;
 
 	D3DDISPLAYMODE d3ddm;
@@ -80,6 +113,11 @@ bool D3D9_Init(HWND hWnd, bool windowed, std::string *error_message) {
 	else
 		dwBehaviorFlags |= D3DCREATE_SOFTWARE_VERTEXPROCESSING;
 
+	RECT rc;
+	GetClientRect(hWnd, &rc);
+	int xres = rc.right - rc.left;
+	int yres = rc.bottom - rc.top;
+
 	D3DPRESENT_PARAMETERS pp;
 	memset(&pp, 0, sizeof(pp));
 	pp.BackBufferWidth = xres;
@@ -93,21 +131,41 @@ bool D3D9_Init(HWND hWnd, bool windowed, std::string *error_message) {
 	pp.AutoDepthStencilFormat = D3DFMT_D24S8;
 	pp.PresentationInterval = (g_Config.bVSync) ? D3DPRESENT_INTERVAL_ONE : D3DPRESENT_INTERVAL_IMMEDIATE;
 
-	hr = d3d->CreateDevice(adapterId, D3DDEVTYPE_HAL, hWnd, dwBehaviorFlags, &pp, &device);
+	if (has9Ex) {
+		if (windowed && IsWin7OrLater()) {
+			// This new flip mode gives higher performance.
+			pp.BackBufferCount = 2;
+			pp.SwapEffect = D3DSWAPEFFECT_FLIPEX;
+		}
+		hr = d3dEx->CreateDeviceEx(adapterId, D3DDEVTYPE_HAL, hWnd, dwBehaviorFlags, &pp, NULL, &deviceEx);
+		device = deviceEx;
+	} else {
+		hr = d3d->CreateDevice(adapterId, D3DDEVTYPE_HAL, hWnd, dwBehaviorFlags, &pp, &device);
+	}
+
 	if (FAILED(hr)) {
 		ELOG("Failed to create D3D device");
-		d3d->Release();
+		if (has9Ex) {
+			d3dEx->Release();
+		} else {
+			d3d->Release();
+		}
 		return false;
 	}
 
 	device->BeginScene();
 	DX9::pD3Ddevice = device;
+	DX9::pD3DdeviceEx = deviceEx;
 
 	LoadD3DX9Dynamic();
 
 	DX9::CompileShaders();
 	DX9::fbo_init();
 
+	if (deviceEx && IsWin7OrLater()) {
+		deviceEx->SetMaximumFrameLatency(1);
+	}
+
 	return true;
 }
 
diff --git a/native b/native
index 3c0d666e90..33d496d76f 160000
--- a/native
+++ b/native
@@ -1 +1 @@
-Subproject commit 3c0d666e90e528a91d76d0d07b312ae76b4d1c63
+Subproject commit 33d496d76f529a6b921c4c8fef70e6c5fd2caef7