diff --git a/plugins/GSdx/GPURenderer.cpp b/plugins/GSdx/GPURenderer.cpp
index 954732b2b1..944d5d7914 100644
--- a/plugins/GSdx/GPURenderer.cpp
+++ b/plugins/GSdx/GPURenderer.cpp
@@ -211,6 +211,6 @@ LRESULT GPURenderer::OnMessage(UINT message, WPARAM wParam, LPARAM lParam)
 		}
 	}
 
-	return m_wndproc(m_hWnd, message, wParam, lParam);
+	return CallWindowProc(m_wndproc, m_hWnd, message, wParam, lParam);
 }
 
diff --git a/plugins/GSdx/GPURendererSW.cpp b/plugins/GSdx/GPURendererSW.cpp
index c71f72fdcb..c4b63199c2 100644
--- a/plugins/GSdx/GPURendererSW.cpp
+++ b/plugins/GSdx/GPURendererSW.cpp
@@ -78,7 +78,7 @@ GSTexture* GPURendererSW::GetOutput()
 		}
 	}
 
-	m_texture->Update(GSVector4i(0, 0, w, h), buff, m_mem.GetWidth() * sizeof(uint32));
+	m_texture->Update(r.rsize(), buff, m_mem.GetWidth() * sizeof(uint32));
 
 	return m_texture;
 }
diff --git a/plugins/GSdx/GS.cpp b/plugins/GSdx/GS.cpp
index bf7ce343c9..f364ba8635 100644
--- a/plugins/GSdx/GS.cpp
+++ b/plugins/GSdx/GS.cpp
@@ -126,7 +126,7 @@ static INT32 GSopen(void* dsp, char* title, int mt, int renderer)
 	case 3: s_gs = new GSRendererHW10(s_basemem, !!mt, s_irq); break;
 	case 4: s_gs = new GSRendererSW(s_basemem, !!mt, s_irq, new GSDevice10()); break;
 	case 5: s_gs = new GSRendererNull(s_basemem, !!mt, s_irq, new GSDevice10()); break;
-	#if 0
+	#if 1
 	case 6: s_gs = new GSRendererOGL(s_basemem, !!mt, s_irq); break;
 	case 7: s_gs = new GSRendererSW(s_basemem, !!mt, s_irq, new GSDeviceOGL()); break;
 	case 8: s_gs = new GSRendererNull(s_basemem, !!mt, s_irq, new GSDeviceOGL()); break;
diff --git a/plugins/GSdx/GSDevice9.cpp b/plugins/GSdx/GSDevice9.cpp
index 3d7dd3a93e..a88a24f97e 100644
--- a/plugins/GSdx/GSDevice9.cpp
+++ b/plugins/GSdx/GSDevice9.cpp
@@ -38,7 +38,6 @@ GSDevice9::GSDevice9()
 	, m_ps_ss(NULL)
 	, m_scissor(0, 0, 0, 0)
 	, m_dss(NULL)
-	, m_sref(0)
 	, m_bs(NULL)
 	, m_bf(0xffffffff)
 	, m_rtv(NULL)
@@ -233,7 +232,6 @@ bool GSDevice9::Reset(int w, int h, bool fs)
 	m_ps_ss = NULL;
 	m_scissor = GSVector4i::zero();
 	m_dss = NULL;
-	m_sref = 0;
 	m_bs = NULL;
 	m_bf = 0xffffffff;
 	m_rtv = NULL;
@@ -570,7 +568,7 @@ void GSDevice9::StretchRect(GSTexture* st, const GSVector4& sr, GSTexture* dt, c
 
 	// om
 
-	OMSetDepthStencilState(&m_convert.dss, 0);
+	OMSetDepthStencilState(&m_convert.dss);
 	OMSetBlendState(bs, 0);
 	OMSetRenderTargets(dt, NULL);
 
@@ -817,6 +815,7 @@ void GSDevice9::PSSetSamplerState(Direct3DSamplerState9* ss)
 {
 	if(ss && m_ps_ss != ss)
 	{
+
 		m_dev->SetSamplerState(0, D3DSAMP_ADDRESSU, ss->AddressU);
 		m_dev->SetSamplerState(0, D3DSAMP_ADDRESSV, ss->AddressV);
 		m_dev->SetSamplerState(1, D3DSAMP_ADDRESSU, ss->AddressU);
@@ -850,9 +849,9 @@ void GSDevice9::RSSet(int width, int height, const GSVector4i* scissor)
 	}
 }
 
-void GSDevice9::OMSetDepthStencilState(Direct3DDepthStencilState9* dss, uint32 sref)
+void GSDevice9::OMSetDepthStencilState(Direct3DDepthStencilState9* dss)
 {
-	if(m_dss != dss || m_sref != sref)
+	if(m_dss != dss)
 	{
 		m_dev->SetRenderState(D3DRS_ZENABLE, dss->DepthEnable);
 		m_dev->SetRenderState(D3DRS_ZWRITEENABLE, dss->DepthWriteMask);
@@ -872,11 +871,10 @@ void GSDevice9::OMSetDepthStencilState(Direct3DDepthStencilState9* dss, uint32 s
 			m_dev->SetRenderState(D3DRS_STENCILPASS, dss->StencilPassOp);
 			m_dev->SetRenderState(D3DRS_STENCILFAIL, dss->StencilFailOp);
 			m_dev->SetRenderState(D3DRS_STENCILZFAIL, dss->StencilDepthFailOp);
-			m_dev->SetRenderState(D3DRS_STENCILREF, sref);
+			m_dev->SetRenderState(D3DRS_STENCILREF, dss->StencilRef);
 		}
 
 		m_dss = dss;
-		m_sref = sref;
 	}
 }
 
diff --git a/plugins/GSdx/GSDevice9.h b/plugins/GSdx/GSDevice9.h
index 304c8ae502..741c4c4296 100644
--- a/plugins/GSdx/GSDevice9.h
+++ b/plugins/GSdx/GSDevice9.h
@@ -44,6 +44,7 @@ struct Direct3DDepthStencilState9
     D3DSTENCILOP StencilDepthFailOp;
     D3DSTENCILOP StencilPassOp;
     D3DCMPFUNC StencilFunc;
+	uint32 StencilRef;
 };
 
 struct Direct3DBlendState9
@@ -75,7 +76,6 @@ private:
 	Direct3DSamplerState9* m_ps_ss;
 	GSVector4i m_scissor;
 	Direct3DDepthStencilState9* m_dss;
-	uint32 m_sref;
 	Direct3DBlendState9* m_bs;
 	uint32 m_bf;
 	IDirect3DSurface9* m_rtv;
@@ -168,7 +168,7 @@ public:
 	void PSSetShader(IDirect3DPixelShader9* ps, const float* ps_cb, int ps_cb_len);
 	void PSSetSamplerState(Direct3DSamplerState9* ss);
 	void RSSet(int width, int height, const GSVector4i* scissor = NULL);
-	void OMSetDepthStencilState(Direct3DDepthStencilState9* dss, uint32 sref);
+	void OMSetDepthStencilState(Direct3DDepthStencilState9* dss);
 	void OMSetBlendState(Direct3DBlendState9* bs, uint32 bf);
 	void OMSetRenderTargets(GSTexture* rt, GSTexture* ds);
 
diff --git a/plugins/GSdx/GSDeviceOGL.cpp b/plugins/GSdx/GSDeviceOGL.cpp
index 8c8e929ed2..fd36ded466 100644
--- a/plugins/GSdx/GSDeviceOGL.cpp
+++ b/plugins/GSdx/GSDeviceOGL.cpp
@@ -29,7 +29,14 @@ GSDeviceOGL::GSDeviceOGL()
 	, m_hGLRC(NULL)
 	, m_vbo(0)
 	, m_fbo(0)
+	, m_context(0)
 	, m_topology(-1)
+	, m_ps_ss(NULL)
+	, m_scissor(0, 0, 0, 0)
+	, m_viewport(0, 0)
+	, m_dss(NULL)
+	, m_bs(NULL)
+	, m_bf(-1)
 	, m_rt((GLuint)-1)
 	, m_ds((GLuint)-1)
 {
@@ -41,6 +48,7 @@ GSDeviceOGL::GSDeviceOGL()
 
 GSDeviceOGL::~GSDeviceOGL()
 {
+	if(m_context) cgDestroyContext(m_context);
 	if(m_vbo) glDeleteBuffers(1, &m_vbo);
 	if(m_fbo) glDeleteFramebuffers(1, &m_fbo);
 	
@@ -100,11 +108,6 @@ bool GSDeviceOGL::Create(GSWnd* wnd, bool vsync)
 		return false;
 	}
 
-	if(WGLEW_EXT_swap_control)
-	{
-		wglSwapIntervalEXT(vsync ? 1 : 0);
-	}
-
 	#endif
 
 	if(glewInit() != GLEW_OK)
@@ -134,6 +137,17 @@ bool GSDeviceOGL::Create(GSWnd* wnd, bool vsync)
 	glBindBuffer(GL_ARRAY_BUFFER, m_vbo); CheckError();
 	// TODO: setup layout?
 
+	m_context = cgCreateContext();
+
+	// cgGLSetDebugMode(CG_FALSE); CheckCgError();
+
+	cgSetParameterSettingMode(m_context, CG_DEFERRED_PARAMETER_SETTING); CheckCgError();
+/*
+	struct {CGprofile vs, gs, ps;} m_profile;
+	m_profile.vs = cgGLGetLatestProfile(CG_GL_VERTEX); CheckCgError();
+	m_profile.gs = cgGLGetLatestProfile(CG_GL_GEOMETRY); CheckCgError();
+	m_profile.ps = cgGLGetLatestProfile(CG_GL_FRAGMENT); CheckCgError();
+*/
 	GSVector4i r = wnd->GetClientRect();
 
 	Reset(r.width(), r.height(), false);
@@ -387,6 +401,111 @@ void GSDeviceOGL::IASetPrimitiveTopology(int topology)
 	m_topology = topology;
 }
 
+void GSDeviceOGL::PSSetSamplerState(SamplerStateOGL* ss)
+{
+	if(ss && m_ps_ss != ss)
+	{
+		glActiveTexture(GL_TEXTURE0);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, ss->wrap.s);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, ss->wrap.t);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, ss->filter);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, ss->filter);
+		
+		glActiveTexture(GL_TEXTURE1);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, ss->wrap.s);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, ss->wrap.t);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, ss->filter);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, ss->filter);
+		
+		glActiveTexture(GL_TEXTURE2);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_POINT);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_POINT);
+		
+		glActiveTexture(GL_TEXTURE3);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_POINT);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_POINT);
+		
+		m_ps_ss = ss;
+	}
+}
+
+void GSDeviceOGL::RSSet(int width, int height, const GSVector4i* scissor)
+{
+	if(m_viewport.x != width || m_viewport.y != height)
+	{
+		glViewport(0, 0, width, height); CheckError();
+
+		m_viewport = GSVector2i(width, height);
+	}
+
+	GSVector4i r = scissor ? *scissor : GSVector4i(0, 0, width, height);
+
+	if(!m_scissor.eq(r))
+	{
+		glScissor(r.left, r.top, r.width(), r.height()); CheckError();
+
+		m_scissor = r;
+	}
+}
+
+void GSDeviceOGL::OMSetDepthStencilState(DepthStencilStateOGL* dss)
+{
+	if(m_dss != dss)
+	{
+		if(dss->depth.enable)
+		{
+			glEnable(GL_DEPTH_TEST); CheckError();
+			glDepthFunc(dss->depth.func); CheckError();
+			glDepthMask(dss->depth.write); CheckError();
+		}
+		else
+		{
+			glDisable(GL_DEPTH_TEST); CheckError();
+		}
+
+		if(dss->stencil.enable)
+		{
+			glEnable(GL_STENCIL_TEST); CheckError();
+			glStencilFunc(dss->stencil.func, dss->stencil.ref, dss->stencil.mask); CheckError();
+			glStencilOp(dss->stencil.sfail, dss->stencil.dpfail, dss->stencil.dppass); CheckError();
+			glStencilMask(dss->stencil.wmask); CheckError();
+		}
+		else
+		{
+			glDisable(GL_STENCIL_TEST); CheckError();
+		}
+
+		m_dss = dss;
+	}
+}
+
+void GSDeviceOGL::OMSetBlendState(BlendStateOGL* bs, float bf)
+{
+	if(m_bs != bs || m_bf != bf)
+	{
+		if(bs->enable)
+		{
+			glEnable(GL_BLEND); CheckError();
+			glBlendEquationSeparate(bs->modeRGB, bs->modeAlpha); CheckError();
+			glBlendFuncSeparate(bs->srcRGB, bs->dstRGB, bs->srcAlpha, bs->dstAlpha); CheckError();
+			glBlendColor(bf, bf, bf, 0); CheckError();
+		}
+		else
+		{
+			glDisable(GL_BLEND); CheckError();
+		}
+		
+		glColorMask(bs->mask.r, bs->mask.g, bs->mask.b, bs->mask.a); CheckError();
+
+		m_bs = bs;
+		m_bf = bf;
+	}
+}
+
 void GSDeviceOGL::OMSetRenderTargets(GSTexture* rt, GSTexture* ds)
 {
 	GLuint rti = 0;
@@ -395,17 +514,17 @@ void GSDeviceOGL::OMSetRenderTargets(GSTexture* rt, GSTexture* ds)
 	if(rt) rti = *(GSTextureOGL*)rt;
 	if(ds) dsi = *(GSTextureOGL*)ds;
 
-	// TODO: if(m_rt != rti)
+	if(m_rt != rti)
 	{
 		glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, rti); CheckError();
 
-		// TODO: m_rt = rti;
+		m_rt = rti;
 	}
 
-	// TODO: if(m_ds != dsi)
+	if(m_ds != dsi)
 	{
 		glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_DEPTH_COMPONENT, dsi); CheckError();
 
-		// TODO: m_ds = dsi;
+		m_ds = dsi;
 	}
 }
diff --git a/plugins/GSdx/GSDeviceOGL.h b/plugins/GSdx/GSDeviceOGL.h
index d95b8916b3..850056120c 100644
--- a/plugins/GSdx/GSDeviceOGL.h
+++ b/plugins/GSdx/GSDeviceOGL.h
@@ -24,6 +24,46 @@
 #include "GSDevice.h"
 #include "GSTextureOGL.h"
 
+struct SamplerStateOGL
+{
+	struct {GLenum s, t;} wrap;
+	GLenum filter;
+};
+
+struct DepthStencilStateOGL
+{
+	struct
+	{
+		bool enable;
+		bool write;
+		GLenum func;
+	} depth;
+
+	struct
+	{
+		bool enable;
+		GLenum func;
+		GLint ref;
+		GLuint mask;
+		GLenum sfail;
+		GLenum dpfail;
+		GLenum dppass;
+		GLuint wmask;
+	} stencil;
+};
+
+struct BlendStateOGL
+{
+	bool enable;
+	GLenum srcRGB;
+	GLenum dstRGB;
+	GLenum srcAlpha;
+	GLenum dstAlpha;
+	GLenum modeRGB;
+	GLenum modeAlpha;
+	union {uint8 r:1, g:1, b:1, a:1;} mask;
+};
+
 class GSDeviceOGL : public GSDevice
 {
 	#ifdef _WINDOWS
@@ -42,11 +82,19 @@ class GSDeviceOGL : public GSDevice
 	} m_vertices;
 
 	int m_topology;
+	SamplerStateOGL* m_ps_ss;
+	GSVector4i m_scissor;
+	GSVector2i m_viewport;
+	DepthStencilStateOGL* m_dss;
+	BlendStateOGL* m_bs;
+	float m_bf;
 	GLuint m_rt;
 	GLuint m_ds;
 
 	//
 
+	CGcontext m_context;
+
 	static void OnStaticCgError(CGcontext ctx, CGerror err, void* p) {((GSDeviceOGL*)p)->OnCgError(ctx, err);}
 	void OnCgError(CGcontext ctx, CGerror err);
 
@@ -89,6 +137,10 @@ public:
 	void IASetInputLayout(); // TODO
 	void IASetPrimitiveTopology(int topology);
 
+	void PSSetSamplerState(SamplerStateOGL* ss);
+	void RSSet(int width, int height, const GSVector4i* scissor);
+	void OMSetDepthStencilState(DepthStencilStateOGL* dss);
+	void OMSetBlendState(BlendStateOGL* bs, float bf);
 	void OMSetRenderTargets(GSTexture* rt, GSTexture* ds);
 
 	static void CheckError() 
@@ -114,4 +166,21 @@ public:
 			printf("%d\n", status);
 		}
 	}
+
+	void CheckCgError()
+	{
+		CGerror error;
+
+		const char* s = cgGetLastErrorString(&error);
+
+		if(error != CG_NO_ERROR)
+		{
+			printf("%s\n", s);
+
+			if(error == CG_COMPILER_ERROR)
+			{
+				printf("%s\n", cgGetLastListing(m_context));
+			}
+		}
+	}
 };
diff --git a/plugins/GSdx/GSLocalMemory.cpp b/plugins/GSdx/GSLocalMemory.cpp
index a13bdef939..8f3313dfe3 100644
--- a/plugins/GSdx/GSLocalMemory.cpp
+++ b/plugins/GSdx/GSLocalMemory.cpp
@@ -72,11 +72,6 @@ int GSLocalMemory::blockOffset4[256];
 
 //
 
-uint32 GSLocalMemory::m_xtbl[1024];
-uint32 GSLocalMemory::m_ytbl[1024]; 
-
-//
-
 GSLocalMemory::psm_t GSLocalMemory::m_psm[64];
 
 //
diff --git a/plugins/GSdx/GSLocalMemory.h b/plugins/GSdx/GSLocalMemory.h
index f690fafa2a..a09e972c8d 100644
--- a/plugins/GSdx/GSLocalMemory.h
+++ b/plugins/GSdx/GSLocalMemory.h
@@ -841,8 +841,6 @@ public:
 
 	//
 
-	static uint32 m_xtbl[1024], m_ytbl[1024]; 
-
 	template<typename T> void ReadTexture(const GSVector4i& r, uint8* dst, int dstpitch, const GIFRegTEX0& TEX0, const GIFRegTEXA& TEXA, readTexel rt, readTexture rtx);
 
 	//
diff --git a/plugins/GSdx/GSRendererHW10.cpp b/plugins/GSdx/GSRendererHW10.cpp
index c8d4dfaa67..39a66640e3 100644
--- a/plugins/GSdx/GSRendererHW10.cpp
+++ b/plugins/GSdx/GSRendererHW10.cpp
@@ -252,17 +252,11 @@ void GSRendererHW10::Draw(int prim, GSTexture* rt, GSTexture* ds, GSTextureCache
 
 	GSTextureFX10::VSSelector vs_sel;
 
-	vs_sel.bpp = 0;
 	vs_sel.bppz = 0;
 	vs_sel.tme = PRIM->TME;
 	vs_sel.fst = PRIM->FST;
 	vs_sel.prim = prim;
 
-	if(tex)
-	{
-		vs_sel.bpp = tex->m_bpp2;
-	}
-
 	if(om_dssel.zte && om_dssel.ztst > 0 && om_dssel.zwe)
 	{
 		if(context->ZBUF.PSM == PSM_PSMZ24)
@@ -349,7 +343,7 @@ void GSRendererHW10::Draw(int prim, GSTexture* rt, GSTexture* ds, GSTextureCache
 
 	if(tex)
 	{
-		ps_sel.bpp = tex->m_bpp2;
+		ps_sel.bpp = tex->m_bpp;
 
 		int w = tex->m_texture->GetWidth();
 		int h = tex->m_texture->GetHeight();
diff --git a/plugins/GSdx/GSRendererHW9.cpp b/plugins/GSdx/GSRendererHW9.cpp
index 5a37df2530..778bb9ef12 100644
--- a/plugins/GSdx/GSRendererHW9.cpp
+++ b/plugins/GSdx/GSRendererHW9.cpp
@@ -50,6 +50,7 @@ bool GSRendererHW9::Create(const string& title)
 	m_date.dss.StencilWriteMask = 1;
 	m_date.dss.StencilFunc = D3DCMP_ALWAYS;
 	m_date.dss.StencilPassOp = D3DSTENCILOP_REPLACE;
+	m_date.dss.StencilRef = 1;
 
 	memset(&m_date.bs, 0, sizeof(m_date.bs));
 
@@ -64,6 +65,7 @@ bool GSRendererHW9::Create(const string& title)
 	m_fba.dss.StencilPassOp = D3DSTENCILOP_ZERO;
 	m_fba.dss.StencilFailOp = D3DSTENCILOP_ZERO;
 	m_fba.dss.StencilDepthFailOp = D3DSTENCILOP_ZERO;
+	m_fba.dss.StencilRef = 2;
 
 	memset(&m_fba.bs, 0, sizeof(m_fba.bs));
 
@@ -321,7 +323,7 @@ void GSRendererHW9::Draw(int prim, GSTexture* rt, GSTexture* ds, GSTextureCache:
 
 	if(tex)
 	{
-		ps_sel.bpp = tex->m_bpp2;
+		ps_sel.bpp = tex->m_bpp;
 
 		int w = tex->m_texture->GetWidth();
 		int h = tex->m_texture->GetHeight();
@@ -523,7 +525,7 @@ void GSRendererHW9::SetupDATE(GSTexture* rt, GSTexture* ds)
 
 		// om
 
-		dev->OMSetDepthStencilState(&m_date.dss, 1);
+		dev->OMSetDepthStencilState(&m_date.dss);
 		dev->OMSetBlendState(&m_date.bs, 0);
 		dev->OMSetRenderTargets(t, ds);
 
@@ -575,7 +577,7 @@ void GSRendererHW9::UpdateFBA(GSTexture* rt)
 
 	// om
 
-	dev->OMSetDepthStencilState(&m_fba.dss, 2);
+	dev->OMSetDepthStencilState(&m_fba.dss);
 	dev->OMSetBlendState(&m_fba.bs, 0);
 
 	// ia
diff --git a/plugins/GSdx/GSTextureCache.cpp b/plugins/GSdx/GSTextureCache.cpp
index 609e9a5667..6bca8d5265 100644
--- a/plugins/GSdx/GSTextureCache.cpp
+++ b/plugins/GSdx/GSTextureCache.cpp
@@ -202,13 +202,14 @@ GSTextureCache::GSCachedTexture* GSTextureCache::GetTexture()
 {
 	const GIFRegTEX0& TEX0 = m_renderer->m_context->TEX0;
 	const GIFRegCLAMP& CLAMP = m_renderer->m_context->CLAMP;
+	const GIFRegTEXA& TEXA = m_renderer->m_env.TEXA;
 
+	const GSLocalMemory::psm_t& psm = GSLocalMemory::m_psm[TEX0.PSM];
 	const uint32* clut = m_renderer->m_mem.m_clut;
-	const int pal = GSLocalMemory::m_psm[TEX0.PSM].pal;
 	
-	if(pal > 0)
+	if(psm.pal > 0)
 	{
-		m_renderer->m_mem.m_clut.Read(TEX0);
+		m_renderer->m_mem.m_clut.Read32(TEX0, TEXA);
 
 		/*
 		POSITION pos = m_tex.GetHeadPosition();
@@ -267,12 +268,17 @@ GSTextureCache::GSCachedTexture* GSTextureCache::GetTexture()
 	{
 		GSCachedTexture* t2 = *i;
 
-		if((((t2->m_TEX0.u32[0] ^ TEX0.u32[0]) & 0xffefffff) | ((t2->m_TEX0.u32[1] ^ TEX0.u32[1]) & 3)) != 0) // TBP0 TBW (PSM & ~1) TW TH
+		if(((t2->m_TEX0.u32[0] ^ TEX0.u32[0]) | ((t2->m_TEX0.u32[1] ^ TEX0.u32[1]) & 3)) != 0) // TBP0 TBW PSM TW TH
 		{
 			continue;
 		}
 
-		if(!(pal == 0 || t2->m_TEX0.CPSM == TEX0.CPSM && GSVector4i::compare(t2->m_clut, clut, pal * sizeof(clut[0]))))
+		if((psm.trbpp == 16 || psm.trbpp == 24) && TEX0.TCC && TEXA != t2->m_TEXA)
+		{
+			continue;
+		}
+
+		if(psm.pal > 0 && !(t2->m_TEX0.CPSM == TEX0.CPSM && GSVector4i::compare(t2->m_clut, clut, psm.pal * sizeof(clut[0]))))
 		{
 			continue;
 		}
@@ -344,23 +350,23 @@ GSTextureCache::GSCachedTexture* GSTextureCache::GetTexture()
 		m_tex.push_front(t);
 	}
 
-	if(pal > 0)
+	if(psm.pal > 0)
 	{
-		int size = pal * sizeof(clut[0]);
+		int size = psm.pal * sizeof(clut[0]);
 
 		if(t->m_palette)
 		{
 			if(t->m_initpalette)
 			{
 				memcpy(t->m_clut, clut, size);
-				t->m_palette->Update(GSVector4i(0, 0, pal, 1), t->m_clut, size);
+				t->m_palette->Update(GSVector4i(0, 0, psm.pal, 1), t->m_clut, size);
 				t->m_initpalette = false;
 			}
 			else
 			{
 				if(GSVector4i::update(t->m_clut, clut, size))
 				{
-					t->m_palette->Update(GSVector4i(0, 0, pal, 1), t->m_clut, size);
+					t->m_palette->Update(GSVector4i(0, 0, psm.pal, 1), t->m_clut, size);
 				}
 			}
 		}
@@ -728,11 +734,11 @@ void GSTextureCache::GSDepthStencil::Update()
 
 GSTextureCache::GSCachedTexture::GSCachedTexture(GSRenderer* r)
 	: GSSurface(r)
-	, m_valid(0, 0, 0, 0)
 	, m_bpp(0)
-	, m_bpp2(0)
 	, m_rendered(false)
 {
+	m_valid = GSVector4i::zero();
+
 	m_clut = (uint32*)_aligned_malloc(256 * sizeof(uint32), 16);
 
 	memset(m_clut, 0, sizeof(m_clut));
@@ -767,7 +773,7 @@ void GSTextureCache::GSCachedTexture::Update()
 	{
 		// in dx9 managed textures can be written directly, less copying is faster, but still not as fast as dx10's UpdateResource
 
-		m_renderer->m_mem.ReadTextureNP(r, m.bits, m.pitch, m_renderer->m_context->TEX0, m_renderer->m_env.TEXA);
+		m_renderer->m_mem.ReadTexture(r, m.bits, m.pitch, m_renderer->m_context->TEX0, m_renderer->m_env.TEXA);
 
 		m_texture->Unmap();
 	}
@@ -777,12 +783,12 @@ void GSTextureCache::GSCachedTexture::Update()
 		
 		int pitch = ((r.width() + 3) & ~3) * 4;
 
-		m_renderer->m_mem.ReadTextureNP(r, buff, pitch, m_renderer->m_context->TEX0, m_renderer->m_env.TEXA);
+		m_renderer->m_mem.ReadTexture(r, buff, pitch, m_renderer->m_context->TEX0, m_renderer->m_env.TEXA);
 
 		m_texture->Update(r, buff, pitch);
 	}
 
-	m_renderer->m_perfmon.Put(GSPerfMon::Unswizzle, r.width() * r.height() * m_bpp >> 3);
+	m_renderer->m_perfmon.Put(GSPerfMon::Unswizzle, r.width() * r.height() * 4);
 }
 
 bool GSTextureCache::GSCachedTexture::GetDirtyRect(GSVector4i& rr)
diff --git a/plugins/GSdx/GSTextureCache.h b/plugins/GSdx/GSTextureCache.h
index a607a18315..ad17e46eb3 100644
--- a/plugins/GSdx/GSTextureCache.h
+++ b/plugins/GSdx/GSTextureCache.h
@@ -38,6 +38,7 @@ public:
 		int m_age;
 		GSDirtyRectList m_dirty;
 		GIFRegTEX0 m_TEX0;
+		GIFRegTEXA m_TEXA;
 
 		explicit GSSurface(GSRenderer* r);
 		virtual ~GSSurface();
@@ -79,7 +80,6 @@ public:
 		uint32* m_clut; // *
 		GSVector4i m_valid;
 		int m_bpp;
-		int m_bpp2;
 		bool m_rendered;
 
 		explicit GSCachedTexture(GSRenderer* renderer);
diff --git a/plugins/GSdx/GSTextureCache10.cpp b/plugins/GSdx/GSTextureCache10.cpp
index 62c6c57b68..2ec9f37db2 100644
--- a/plugins/GSdx/GSTextureCache10.cpp
+++ b/plugins/GSdx/GSTextureCache10.cpp
@@ -129,50 +129,13 @@ bool GSTextureCache10::GSCachedTextureHW10::Create()
 	// m_renderer->m_perfmon.Put(GSPerfMon::WriteTexture, 1);
 
 	m_TEX0 = m_renderer->m_context->TEX0;
+	m_TEXA = m_renderer->m_env.TEXA;
 
-	uint32 psm = m_TEX0.PSM;
-
-	switch(psm)
-	{
-	case PSM_PSMT8:
-	case PSM_PSMT8H:
-	case PSM_PSMT4:
-	case PSM_PSMT4HL:
-	case PSM_PSMT4HH:
-		psm = m_TEX0.CPSM;
-		break;
-	}
-
-	DXGI_FORMAT format;
-
-	switch(psm)
-	{
-	default:
-		// printf("Invalid TEX0.PSM/CPSM (%I64d, %I64d)\n", m_TEX0.PSM, m_TEX0.CPSM);
-	case PSM_PSMCT32:
-		m_bpp = 32;
-		m_bpp2 = 0;
-		format = DXGI_FORMAT_R8G8B8A8_UNORM;
-		break;
-	case PSM_PSMCT24:
-		m_bpp = 32;
-		m_bpp2 = 1;
-		format = DXGI_FORMAT_R8G8B8A8_UNORM;
-		break;
-	case PSM_PSMCT16:
-	case PSM_PSMCT16S:
-		m_bpp = 16;
-		m_bpp2 = 5;
-		format = DXGI_FORMAT_R16_UNORM;
-		break;
-	}
-
-	int w = 1 << m_TEX0.TW;
-	int h = 1 << m_TEX0.TH;
+	m_bpp = 0;
 
 	ASSERT(m_texture == NULL);
 
-	m_texture = m_renderer->m_dev->CreateTexture(w, h, format);
+	m_texture = m_renderer->m_dev->CreateTexture(1 << m_TEX0.TW, 1 << m_TEX0.TH);
 
 	return m_texture != NULL;
 }
@@ -186,6 +149,8 @@ bool GSTextureCache10::GSCachedTextureHW10::Create(GSRenderTarget* rt)
 	// m_renderer->m_perfmon.Put(GSPerfMon::ConvertRT2T, 1);
 
 	m_TEX0 = m_renderer->m_context->TEX0;
+	m_TEXA = m_renderer->m_env.TEXA;
+
 	m_rendered = true;
 
 	int tw = 1 << m_TEX0.TW;
@@ -309,18 +274,18 @@ bool GSTextureCache10::GSCachedTextureHW10::Create(GSRenderTarget* rt)
 	switch(m_TEX0.PSM)
 	{
 	case PSM_PSMCT32:
-		m_bpp2 = 0;
+		m_bpp = 0;
 		break;
 	case PSM_PSMCT24:
-		m_bpp2 = 1;
+		m_bpp = 1;
 		break;
 	case PSM_PSMCT16:
 	case PSM_PSMCT16S:
-		m_bpp2 = 2;
+		m_bpp = 2;
 		break;
 	case PSM_PSMT8H:
-		m_bpp2 = 3;
-		m_palette = m_renderer->m_dev->CreateTexture(256, 1, m_TEX0.CPSM == PSM_PSMCT32 ? DXGI_FORMAT_R8G8B8A8_UNORM : DXGI_FORMAT_R16_UNORM); // 
+		m_bpp = 3;
+		m_palette = m_renderer->m_dev->CreateTexture(256, 1);
 		m_initpalette = true;
 		break;
 	case PSM_PSMT4HL:
diff --git a/plugins/GSdx/GSTextureCache9.cpp b/plugins/GSdx/GSTextureCache9.cpp
index d288c5ebac..e9b685ae17 100644
--- a/plugins/GSdx/GSTextureCache9.cpp
+++ b/plugins/GSdx/GSTextureCache9.cpp
@@ -127,50 +127,13 @@ bool GSTextureCache9::GSCachedTexture9::Create()
 	// m_renderer->m_perfmon.Put(GSPerfMon::WriteTexture, 1);
 
 	m_TEX0 = m_renderer->m_context->TEX0;
+	m_TEXA = m_renderer->m_env.TEXA;
 
-	uint32 psm = m_TEX0.PSM;
-
-	switch(psm)
-	{
-	case PSM_PSMT8:
-	case PSM_PSMT8H:
-	case PSM_PSMT4:
-	case PSM_PSMT4HL:
-	case PSM_PSMT4HH:
-		psm = m_TEX0.CPSM;
-		break;
-	}
-
-	D3DFORMAT format;
-
-	switch(psm)
-	{
-	default:
-		// printf("Invalid TEX0.PSM/CPSM (%I64d, %I64d)\n", m_TEX0.PSM, m_TEX0.CPSM);
-	case PSM_PSMCT32:
-		m_bpp = 32;
-		m_bpp2 = 0;
-		format = D3DFMT_A8R8G8B8;
-		break;
-	case PSM_PSMCT24:
-		m_bpp = 32;
-		m_bpp2 = 1;
-		format = D3DFMT_A8R8G8B8;
-		break;
-	case PSM_PSMCT16:
-	case PSM_PSMCT16S:
-		m_bpp = 16;
-		m_bpp2 = 2;
-		format = D3DFMT_A1R5G5B5;
-		break;
-	}
-
-	int w = 1 << m_TEX0.TW;
-	int h = 1 << m_TEX0.TH;
+	m_bpp = 0;
 
 	ASSERT(m_texture == NULL);
 
-	m_texture = m_renderer->m_dev->CreateTexture(w, h, format);
+	m_texture = m_renderer->m_dev->CreateTexture(1 << m_TEX0.TW, 1 << m_TEX0.TH);
 
 	return m_texture != NULL;
 }
@@ -184,6 +147,8 @@ bool GSTextureCache9::GSCachedTexture9::Create(GSRenderTarget* rt)
 	// m_renderer->m_perfmon.Put(GSPerfMon::ConvertRT2T, 1);
 
 	m_TEX0 = m_renderer->m_context->TEX0;
+	m_TEXA = m_renderer->m_env.TEXA;
+
 	m_rendered = true;
 
 	int tw = 1 << m_TEX0.TW;
@@ -307,18 +272,18 @@ bool GSTextureCache9::GSCachedTexture9::Create(GSRenderTarget* rt)
 	switch(m_TEX0.PSM)
 	{
 	case PSM_PSMCT32:
-		m_bpp2 = 0;
+		m_bpp = 0;
 		break;
 	case PSM_PSMCT24:
-		m_bpp2 = 1;
+		m_bpp = 1;
 		break;
 	case PSM_PSMCT16:
 	case PSM_PSMCT16S:
-		m_bpp2 = 2;
+		m_bpp = 2;
 		break;
 	case PSM_PSMT8H:
-		m_bpp2 = 3;
-		m_palette = m_renderer->m_dev->CreateTexture(256, 1, m_TEX0.CPSM == PSM_PSMCT32 ? D3DFMT_A8R8G8B8 : D3DFMT_A1R5G5B5);
+		m_bpp = 3;
+		m_palette = m_renderer->m_dev->CreateTexture(256, 1);
 		m_initpalette = true;
 		break;
 	case PSM_PSMT4HL:
diff --git a/plugins/GSdx/GSTextureCacheSW.cpp b/plugins/GSdx/GSTextureCacheSW.cpp
index ca7144a438..1be3e4b6a6 100644
--- a/plugins/GSdx/GSTextureCacheSW.cpp
+++ b/plugins/GSdx/GSTextureCacheSW.cpp
@@ -51,7 +51,7 @@ const GSTextureCacheSW::GSTexture* GSTextureCacheSW::Lookup(const GIFRegTEX0& TE
 			continue;
 		}
 
-		if((psm.trbpp == 16 || psm.trbpp == 24) && (t2->m_TEX0.TCC != TEX0.TCC || TEX0.TCC && TEXA != t2->m_TEXA))
+		if((psm.trbpp == 16 || psm.trbpp == 24) && TEX0.TCC && TEXA != t2->m_TEXA)
 		{
 			continue;
 		}
diff --git a/plugins/GSdx/GSTextureFX10.cpp b/plugins/GSdx/GSTextureFX10.cpp
index b00de4e1d0..a3b04e1cc9 100644
--- a/plugins/GSdx/GSTextureFX10.cpp
+++ b/plugins/GSdx/GSTextureFX10.cpp
@@ -104,21 +104,19 @@ bool GSTextureFX10::SetupVS(VSSelector sel, const VSConstantBuffer* cb)
 
 	if(i == m_vs.end())
 	{
-		string str[5];
+		string str[4];
 
-		str[0] = format("%d", sel.bpp);
-		str[1] = format("%d", sel.bppz);
-		str[2] = format("%d", sel.tme);
-		str[3] = format("%d", sel.fst);
-		str[4] = format("%d", sel.prim);
+		str[0] = format("%d", sel.bppz);
+		str[1] = format("%d", sel.tme);
+		str[2] = format("%d", sel.fst);
+		str[3] = format("%d", sel.prim);
 
 		D3D10_SHADER_MACRO macro[] =
 		{
-			{"VS_BPP", str[0].c_str()},
-			{"VS_BPPZ", str[1].c_str()},
-			{"VS_TME", str[2].c_str()},
-			{"VS_FST", str[3].c_str()},
-			{"VS_PRIM", str[4].c_str()},
+			{"VS_BPPZ", str[0].c_str()},
+			{"VS_TME", str[1].c_str()},
+			{"VS_FST", str[2].c_str()},
+			{"VS_PRIM", str[3].c_str()},
 			{NULL, NULL},
 		};
 
diff --git a/plugins/GSdx/GSTextureFX10.h b/plugins/GSdx/GSTextureFX10.h
index f50617052b..285d930538 100644
--- a/plugins/GSdx/GSTextureFX10.h
+++ b/plugins/GSdx/GSTextureFX10.h
@@ -63,7 +63,6 @@ public:
 	{
 		struct
 		{
-			uint32 bpp:3;
 			uint32 bppz:2;
 			uint32 tme:1;
 			uint32 fst:1;
@@ -72,7 +71,7 @@ public:
 
 		uint32 key;
 
-		operator uint32() {return key & 0x3ff;}
+		operator uint32() {return key & 0x7f;}
 	};
 
 	__declspec(align(16)) struct PSConstantBuffer
diff --git a/plugins/GSdx/GSTextureFX9.cpp b/plugins/GSdx/GSTextureFX9.cpp
index f59d70246f..c303db9652 100644
--- a/plugins/GSdx/GSTextureFX9.cpp
+++ b/plugins/GSdx/GSTextureFX9.cpp
@@ -293,6 +293,7 @@ void GSTextureFX9::UpdateOM(OMDepthStencilSelector dssel, OMBlendSelector bsel,
 			dss->StencilPassOp = dssel.fba ? D3DSTENCILOP_REPLACE : D3DSTENCILOP_KEEP;
 			dss->StencilFailOp = dssel.fba ? D3DSTENCILOP_ZERO : D3DSTENCILOP_KEEP;
 			dss->StencilDepthFailOp = dssel.fba ? D3DSTENCILOP_ZERO : D3DSTENCILOP_KEEP;
+			dss->StencilRef = 3;
 		}
 
 		if(!(dssel.zte && dssel.ztst == 1 && !dssel.zwe))
@@ -315,7 +316,7 @@ void GSTextureFX9::UpdateOM(OMDepthStencilSelector dssel, OMBlendSelector bsel,
 		i = m_om_dss.find(dssel);
 	}
 
-	m_dev->OMSetDepthStencilState((*i).second, 3);
+	m_dev->OMSetDepthStencilState((*i).second);
 
 	hash_map<uint32, Direct3DBlendState9*>::const_iterator j = m_om_bs.find(bsel);
 
diff --git a/plugins/GSdx/res/tfx10.fx b/plugins/GSdx/res/tfx10.fx
index a2b067a93b..529e528ab7 100644
--- a/plugins/GSdx/res/tfx10.fx
+++ b/plugins/GSdx/res/tfx10.fx
@@ -22,8 +22,7 @@ struct VS_OUTPUT
 	float4 c : COLOR0;
 };
 
-#ifndef VS_BPP
-#define VS_BPP 0
+#ifndef VS_BPPZ
 #define VS_BPPZ 0
 #define VS_TME 1
 #define VS_FST 1
@@ -193,23 +192,6 @@ struct PS_OUTPUT
 #define LTF 1
 #endif
 
-float4 Normalize16(float4 f)
-{
-	return f / float4(0x001f, 0x03e0, 0x7c00, 0x8000);
-}
-
-float4 Extract16(uint i)
-{
-	float4 f;
-
-	f.r = i & 0x001f;
-	f.g = i & 0x03e0;
-	f.b = i & 0x7c00;
-	f.a = i & 0x8000;
-
-	return f;
-}
-
 float4 wrapuv(float4 uv)
 {
 	if(WMS == WMT)
@@ -318,7 +300,7 @@ float4 sample(float2 tc, float w)
 
 		float4 t00, t01, t10, t11;
 
-		if(BPP == 3) // 8HP + 32-bit palette
+		if(BPP == 3) // 8HP
 		{
 			float4 a;
 
@@ -332,26 +314,6 @@ float4 sample(float2 tc, float w)
 			t10 = Palette.Sample(PaletteSampler, a.z);
 			t11 = Palette.Sample(PaletteSampler, a.w);
 		}
-		else if(BPP == 4) // 8HP + 16-bit palette
-		{
-			// TODO: yuck, just pre-convert the palette to 32-bit
-		}
-		else if(BPP == 5) // 16P
-		{
-			float4 r;
-
-			r.x = Texture.Sample(TextureSampler, uv.xy).r;
-			r.y = Texture.Sample(TextureSampler, uv.zy).r;
-			r.z = Texture.Sample(TextureSampler, uv.xw).r;
-			r.w = Texture.Sample(TextureSampler, uv.zw).r;
-			
-			uint4 i = r * 65535;
-
-			t00 = Extract16(i.x);
-			t01 = Extract16(i.y);
-			t10 = Extract16(i.z);
-			t11 = Extract16(i.w);
-		}
 		else
 		{
 			t00 = Texture.Sample(TextureSampler, uv.xy);
@@ -374,13 +336,8 @@ float4 sample(float2 tc, float w)
 	{
 		t.a = AEM == 0 || any(t.rgb) ? TA.x : 0;
 	}
-	else if(BPP == 2 || BPP == 5) // 16 || 16P
+	else if(BPP == 2) // 16
 	{
-		if(BPP == 5)
-		{
-			t = Normalize16(t);
-		}
-		
 		// a bit incompatible with up-scaling because the 1 bit alpha is interpolated
 		
 		t.a = t.a >= 0.5 ? TA.y : AEM == 0 || any(t.rgb) ? TA.x : 0;
diff --git a/plugins/GSdx/res/tfx9.fx b/plugins/GSdx/res/tfx9.fx
index 1cbc591387..c7a2c57896 100644
--- a/plugins/GSdx/res/tfx9.fx
+++ b/plugins/GSdx/res/tfx9.fx
@@ -219,7 +219,7 @@ float4 sample(float2 tc, float w)
 
 		float4 t00, t01, t10, t11;
 
-		if(BPP == 3) // 8HP ln
+		if(BPP == 3) // 8HP
 		{
 			float4 a;