diff --git a/plugins/GSdx/GS.cpp b/plugins/GSdx/GS.cpp
index 582556ae8e..8c448348fc 100644
--- a/plugins/GSdx/GS.cpp
+++ b/plugins/GSdx/GS.cpp
@@ -246,7 +246,7 @@ static int _GSopen(void** dsp, char* title, int renderer, int threads = -1)
 				case 2: dev = new GSDeviceSDL(); break;
 				#endif
 				case 3: dev = new GSDeviceNull(); break;
-				case 5: dev = new GSDeviceOGL(); break;
+				case 4: dev = new GSDeviceOGL(); break;
 			}
 
 			if(dev == NULL)
@@ -429,11 +429,15 @@ EXPORT_C GSreadFIFO(uint8* mem)
 #ifdef _LINUX
 	// FIXME: double check which thread call this function
 	// See fifo2 issue below
-	if (theApp.GetConfig("renderer", 0) == 12) {
+	if (theApp.GetConfig("renderer", 0) / 3 == 4) {
 		fprintf(stderr, "Disable FIFO1 on opengl\n");
 	}
 #endif
+	s_gs->m_wnd.AttachContext();
+
 	s_gs->ReadFIFO(mem, 1);
+
+	s_gs->m_wnd.DetachContext();
 }
 
 EXPORT_C GSreadFIFO2(uint8* mem, uint32 size)
@@ -441,14 +445,18 @@ EXPORT_C GSreadFIFO2(uint8* mem, uint32 size)
 #ifdef _LINUX
 	// FIXME called from EE core thread not MTGS which cause
 	// invalidate data for opengl
-	if (theApp.GetConfig("renderer", 0) == 12) {
+	if (theApp.GetConfig("renderer", 0) / 3 == 4) {
 #ifdef OGL_DEBUG
 		fprintf(stderr, "Disable FIFO2(%d) on opengl\n", size);
 #endif
 		//return;
 	}
 #endif
+	s_gs->m_wnd.AttachContext();
+
 	s_gs->ReadFIFO(mem, size);
+
+	s_gs->m_wnd.DetachContext();
 }
 
 EXPORT_C GSgifTransfer(const uint8* mem, uint32 size)
diff --git a/plugins/GSdx/GSDeviceOGL.cpp b/plugins/GSdx/GSDeviceOGL.cpp
index 2d7ef96c53..63a5078d41 100644
--- a/plugins/GSdx/GSDeviceOGL.cpp
+++ b/plugins/GSdx/GSDeviceOGL.cpp
@@ -536,6 +536,7 @@ void GSDeviceOGL::DebugOutput()
 		fprintf(stderr, "\n");
 
 	}
+	CheckDebugLog();
 }
 
 void GSDeviceOGL::DrawPrimitive()
@@ -566,6 +567,20 @@ void GSDeviceOGL::DrawIndexedPrimitive()
 #endif
 }
 
+void GSDeviceOGL::DrawIndexedPrimitive(int offset, int count)
+{
+	ASSERT(offset + count <= m_index.count);
+#ifdef OGL_DEBUG
+	DebugInput();
+#endif
+
+	m_state.vb->DrawIndexedPrimitive(offset, count);
+#ifdef OGL_DEBUG
+	DebugOutput();
+	g_draw_count++;
+#endif
+}
+
 void GSDeviceOGL::ClearRenderTarget(GSTexture* t, const GSVector4& c)
 {
 	GLuint fbo_old = m_state.fbo;
@@ -578,9 +593,15 @@ void GSDeviceOGL::ClearRenderTarget(GSTexture* t, const GSVector4& c)
 		// glClearColor(c.x, c.y, c.z, c.w);
 		// glClear(GL_COLOR_BUFFER_BIT);
 	} else {
-		// FIXME I need to clarify this FBO attachment stuff
+		// FIXME1 I need to clarify this FBO attachment stuff
 		// I would like to avoid FBO for a basic clean operation
 		OMSetFBO(m_fbo);
+		// FIXME2:
+		// it would be better to attach the texture to another slot (like 3)
+		// but it crash for an unknow reason
+		// The following error appears "glClearBufferfv failed because the currently set GL_DRAW_FRAMEBUFFER binding has incomplete status"
+		// Maybe glDrawBuffer must be updated but it does not solve the crash...
+		// FIXME3: I still have another crash on GOW...
 		static_cast<GSTextureOGL*>(t)->Attach(GL_COLOR_ATTACHMENT0);
 		glClearBufferfv(GL_COLOR, 0, c.v);
 	}
@@ -970,6 +991,8 @@ void GSDeviceOGL::SetUniformBuffer(GSUniformBufferOGL* cb)
 
 void GSDeviceOGL::IASetVertexState(GSVertexBufferStateOGL* vb)
 {
+	if (vb == NULL) vb = m_vb;
+
 	if (m_state.vb != vb) {
 		m_state.vb = vb;
 		vb->bind();
@@ -981,6 +1004,16 @@ void GSDeviceOGL::IASetVertexBuffer(const void* vertices, size_t count)
 	m_state.vb->UploadVB(vertices, count);
 }
 
+bool GSDeviceOGL::IAMapVertexBuffer(void** vertex, size_t stride, size_t count)
+{
+	return m_state.vb->MapVB(vertex, count);
+}
+
+void GSDeviceOGL::IAUnmapVertexBuffer()
+{
+	m_state.vb->UnmapVB();
+}
+
 void GSDeviceOGL::IASetIndexBuffer(const void* index, size_t count)
 {
 	m_state.vb->UploadIB(index, count);
@@ -1073,6 +1106,8 @@ void GSDeviceOGL::OMSetFBO(GLuint fbo)
 	if (m_state.fbo != fbo) {
 		m_state.fbo = fbo;
 		glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+		// FIXME DEBUG
+		//if (fbo) fprintf(stderr, "FB status %x\n", glCheckFramebufferStatus(GL_FRAMEBUFFER));
 	}
 }
 
@@ -1277,6 +1312,7 @@ void GSDeviceOGL::CheckDebugLog()
 void GSDeviceOGL::DebugOutputToFile(unsigned int source, unsigned int type, unsigned int id, unsigned int severity, const char* message)
 {
 	char debType[20], debSev[5];
+	static int sev_counter = 0;
 
 	if(type == GL_DEBUG_TYPE_ERROR_ARB)
 		strcpy(debType, "Error");
@@ -1293,23 +1329,26 @@ void GSDeviceOGL::DebugOutputToFile(unsigned int source, unsigned int type, unsi
 	else
 		strcpy(debType, "UNKNOWN");
 
-	if(severity == GL_DEBUG_SEVERITY_HIGH_ARB)
+	if(severity == GL_DEBUG_SEVERITY_HIGH_ARB) {
 		strcpy(debSev, "High");
+		sev_counter++;
+	}
 	else if(severity == GL_DEBUG_SEVERITY_MEDIUM_ARB)
 		strcpy(debSev, "Med");
 	else if(severity == GL_DEBUG_SEVERITY_LOW_ARB)
 		strcpy(debSev, "Low");
 
 	#ifdef LOUD_DEBUGGING
-	fprintf(stderr,"Type:%s\tID:%d\tSeverity:%s\tMessage:%s\n", debType,id,debSev,message);
+	fprintf(stderr,"Type:%s\tID:%d\tSeverity:%s\tMessage:%s\n", debType, g_draw_count, debSev,message);
 	#endif
 
 	FILE* f = fopen("Debug.txt","a");
 	if(f)
 	{
-		fprintf(f,"Type:%s\tID:%d\tSeverity:%s\tMessage:%s\n", debType,id,debSev,message);
+		fprintf(f,"Type:%s\tID:%d\tSeverity:%s\tMessage:%s\n", debType, g_draw_count, debSev,message);
 		fclose(f);
 	}
+	//if (sev_counter > 2) assert(0);
 }
 
 // (A - B) * C + D
diff --git a/plugins/GSdx/GSDeviceOGL.h b/plugins/GSdx/GSDeviceOGL.h
index 21121f7403..fc574a5229 100644
--- a/plugins/GSdx/GSDeviceOGL.h
+++ b/plugins/GSdx/GSDeviceOGL.h
@@ -323,11 +323,22 @@ class GSVertexBufferStateOGL {
 
 		void upload(const void* src, uint32 count)
 		{
+			// Upload the data to the buffer
+			void* dst;
+			if (Map(&dst, count)) {
+				// FIXME which one to use
+				// GSVector4i::storent(dst, src, m_count * m_stride);
+				memcpy(dst, src, m_stride*m_count);
+				Unmap();
+			}
+		}
+
+		bool Map(void** pointer, uint32 count ) {
 #ifdef OGL_DEBUG
 			GLint b_size = -1;
 			glGetBufferParameteriv(m_target, GL_BUFFER_SIZE, &b_size);
 
-			if (b_size <= 0) return;
+			if (b_size <= 0) return false;
 #endif
 
 			m_count = count;
@@ -338,7 +349,7 @@ class GSVertexBufferStateOGL {
 
 			// Current GPU buffer is really too small need to allocate a new one
 			if (m_count > m_limit) {
-				allocate(std::max<int>(count * 3 / 2, m_default_size));
+				allocate(std::max<int>(m_count * 3 / 2, m_default_size));
 
 			} else if (m_count > (m_limit - m_start) ) {
 				// Not enough left free room. Just go back at the beginning
@@ -354,17 +365,19 @@ class GSVertexBufferStateOGL {
 			}
 
 			// Upload the data to the buffer
-			uint8* dst = (uint8*) glMapBufferRange(m_target, m_stride*m_start, m_stride*m_count, map_flags);
+			*pointer = (uint8*) glMapBufferRange(m_target, m_stride*m_start, m_stride*m_count, map_flags);
+			//fprintf(stderr, "Map %x from %d to %d\n", *pointer, m_start, m_start+m_count);
 #ifdef OGL_DEBUG
-			if (dst == NULL) {
+			if (*pointer == NULL) {
 				fprintf(stderr, "CRITICAL ERROR map failed for vb!!!\n");
-				return;
+				return false;
 			}
 #endif
-			memcpy(dst, src, m_stride*m_count);
-			glUnmapBuffer(m_target);
+			return true;
 		}
 
+		void Unmap() { glUnmapBuffer(m_target); }
+
 		void EndScene()
 		{
 			m_start += m_count;
@@ -381,6 +394,11 @@ class GSVertexBufferStateOGL {
 			glDrawElementsBaseVertex(mode, m_count, GL_UNSIGNED_INT, (void*)(m_start * m_stride), basevertex);
 		}
 
+		void Draw(GLenum mode, GLint basevertex, int offset, int count)
+		{
+			glDrawElementsBaseVertex(mode, count, GL_UNSIGNED_INT, (void*)((m_start + offset) * m_stride), basevertex);
+		}
+
 		size_t GetStart() { return m_start; }
 
 		void debug() 
@@ -416,52 +434,6 @@ public:
 		m_vb->bind();
 	}
 
-#if 0
-	void upload(const void* src, uint32 count)
-	{
-#ifdef OGL_DEBUG
-		GLint b_size = -1;
-		glGetBufferParameteriv(m_target, GL_BUFFER_SIZE, &b_size);
-		
-		if (b_size <= 0) return;
-#endif
-		
-		m_count = count;
-
-		// Note: For an explanation of the map flag
-		// see http://www.opengl.org/wiki/Buffer_Object_Streaming
-		uint32 map_flags = GL_MAP_WRITE_BIT | GL_MAP_UNSYNCHRONIZED_BIT;
-
-		// Current GPU buffer is really too small need to allocate a new one
-		if (m_count > m_limit) {
-			allocate(std::max<int>(count * 3 / 2, 60000));
-
-		} else if (m_count > (m_limit - m_start) ) {
-			// Not enough left free room. Just go back at the beginning
-			m_start = 0;
-
-			// Tell the driver that it can orphan previous buffer and restart from a scratch buffer.
-			// Technically the buffer will not be accessible by the application anymore but the
-			// GL will effectively remove it when draws call are finised.
-			map_flags |= GL_MAP_INVALIDATE_BUFFER_BIT;
-		} else {
-			// Tell the driver that it doesn't need to contain any valid buffer data, and that you promise to write the entire range you map
-			map_flags |= GL_MAP_INVALIDATE_RANGE_BIT;
-		}
-		
-		// Upload the data to the buffer
-		uint8* dst = (uint8*) glMapBufferRange(m_target, m_stride*m_start, m_stride*m_count, map_flags);
-#ifdef OGL_DEBUG
-		if (dst == NULL) {
-			fprintf(stderr, "CRITICAL ERROR map failed for vb!!!\n");
-			return;
-		}
-#endif
-		memcpy(dst, src, m_stride*m_count);
-		glUnmapBuffer(m_target);
-	}
-#endif
-
 	void set_internal_format(GSInputLayoutOGL* layout, uint32 layout_nbr)
 	{
 		for (int i = 0; i < layout_nbr; i++) {
@@ -490,17 +462,17 @@ public:
 
 	void DrawIndexedPrimitive() { m_ib->Draw(m_topology, m_vb->GetStart() ); }
 
+	void DrawIndexedPrimitive(int offset, int count) { m_ib->Draw(m_topology, m_vb->GetStart(), offset, count ); }
+
 	void SetTopology(GLenum topology) { m_topology = topology; }
 
-	void UploadVB(const void* vertices, size_t count)
-	{
-		m_vb->upload(vertices, count);
-	}
+	void UploadVB(const void* vertices, size_t count) { m_vb->upload(vertices, count); }
 
-	void UploadIB(const void* index, size_t count)
-	{
-		m_ib->upload(index, count);
-	}
+	void UploadIB(const void* index, size_t count) { m_ib->upload(index, count); }
+
+	bool MapVB(void **pointer, size_t count) { return m_vb->Map(pointer, count); }
+
+	void UnmapVB() { m_vb->Unmap(); }
 
 	~GSVertexBufferStateOGL()
 	{
@@ -878,6 +850,7 @@ class GSDeviceOGL : public GSDevice
 
 	void DrawPrimitive();
 	void DrawIndexedPrimitive();
+	void DrawIndexedPrimitive(int offset, int count);
 
 	void ClearRenderTarget(GSTexture* t, const GSVector4& c);
 	void ClearRenderTarget(GSTexture* t, uint32 c);
@@ -906,8 +879,10 @@ class GSDeviceOGL : public GSDevice
 
 	void IASetPrimitiveTopology(GLenum topology);
 	void IASetVertexBuffer(const void* vertices, size_t count);
+	bool IAMapVertexBuffer(void** vertex, size_t stride, size_t count);
+	void IAUnmapVertexBuffer();
 	void IASetIndexBuffer(const void* index, size_t count);
-	void IASetVertexState(GSVertexBufferStateOGL* vb);
+	void IASetVertexState(GSVertexBufferStateOGL* vb = NULL);
 
 	void SetUniformBuffer(GSUniformBufferOGL* cb);
 
diff --git a/plugins/GSdx/GSLinuxDialog.cpp b/plugins/GSdx/GSLinuxDialog.cpp
index 97b4a2648c..4a0ccd57ab 100644
--- a/plugins/GSdx/GSLinuxDialog.cpp
+++ b/plugins/GSdx/GSLinuxDialog.cpp
@@ -76,8 +76,8 @@ GtkWidget* CreateRenderComboBox()
 		case 8 : renderer_box_position = 1; break;
 		case 10: renderer_box_position = 2; break;
 		case 11: renderer_box_position = 3; break;
-		case 15: renderer_box_position = 4; break;
-		case 16: renderer_box_position = 5; break;
+		case 12: renderer_box_position = 4; break;
+		case 13: renderer_box_position = 5; break;
 	}
 	gtk_combo_box_set_active(GTK_COMBO_BOX(render_combo_box), renderer_box_position);
 	return render_combo_box;
@@ -375,8 +375,8 @@ bool RunLinuxDialog()
 				case 1: theApp.SetConfig("renderer", 8); break;
 				case 2: theApp.SetConfig("renderer", 10); break;
 				case 3: theApp.SetConfig("renderer", 11); break;
-				case 4: theApp.SetConfig("renderer", 15); break;
-				case 5: theApp.SetConfig("renderer", 16); break;
+				case 4: theApp.SetConfig("renderer", 12); break;
+				case 5: theApp.SetConfig("renderer", 13); break;
 			}
 		}
 
diff --git a/plugins/GSdx/GSRendererOGL.cpp b/plugins/GSdx/GSRendererOGL.cpp
index 1a86704ff2..418aceba61 100644
--- a/plugins/GSdx/GSRendererOGL.cpp
+++ b/plugins/GSdx/GSRendererOGL.cpp
@@ -24,15 +24,12 @@
 
 
 GSRendererOGL::GSRendererOGL()
-	: GSRendererHW(new GSVertexTraceDX11(this), sizeof(GSVertexHW11), new GSTextureCacheOGL(this))
-	  , m_topology(0)
+	: GSRendererHW(new GSTextureCacheOGL(this))
 {
 	m_logz = !!theApp.GetConfig("logz", 0);
 	m_fba = !!theApp.GetConfig("fba", 1);
 	UserHacks_AlphaHack = !!theApp.GetConfig("UserHacks_AlphaHack", 0);
 	m_pixelcenter = GSVector2(-0.5f, -0.5f);
-
-	InitConvertVertex(GSRendererOGL);
 }
 
 bool GSRendererOGL::CreateDevice(GSDevice* dev)
@@ -43,44 +40,47 @@ bool GSRendererOGL::CreateDevice(GSDevice* dev)
 	return true;
 }
 
-template<uint32 prim, uint32 tme, uint32 fst>
-void GSRendererOGL::ConvertVertex(size_t dst_index, size_t src_index)
+void GSRendererOGL::SetupIA()
 {
-	GSVertex* s = (GSVertex*)((GSVertexHW11*)m_vertex.buff + src_index);
-	GSVertexHW11* d = (GSVertexHW11*)m_vertex.buff + dst_index;
+	GSDeviceOGL* dev = (GSDeviceOGL*)m_dev;
 
-	GSVector4i v0 = ((GSVector4i*)s)[0];
-	GSVector4i v1 = ((GSVector4i*)s)[1];
+	void* ptr = NULL;
 
-	if(tme && fst)
+	dev->IASetVertexState();
+
+	if(dev->IAMapVertexBuffer(&ptr, sizeof(GSVertex), m_vertex.next))
 	{
-		// TODO: modify VertexTrace and the shaders to read uv from v1.u16[0], v1.u16[1], then this step is not needed
+		GSVector4i::storent(ptr, m_vertex.buff, sizeof(GSVertex) * m_vertex.next);
 
-		v0 = GSVector4i::cast(GSVector4(v1.uph16()).xyzw(GSVector4::cast(v0))); // uv => st
+		dev->IAUnmapVertexBuffer();
 	}
 
-	((GSVector4i*)d)[0] = v0;
-	((GSVector4i*)d)[1] = v1;
-}
+	dev->IASetIndexBuffer(m_index.buff, m_index.tail);
 
-void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Source* tex)
-{
-	switch(m_vt->m_primclass)
+	GLenum t;
+
+	switch(m_vt.m_primclass)
 	{
 	case GS_POINT_CLASS:
-		m_topology = GL_POINTS;
+		t = GL_POINTS;
 		break;
 	case GS_LINE_CLASS:
 	case GS_SPRITE_CLASS:
-		m_topology = GL_LINES;
+		t = GL_LINES;
 		break;
 	case GS_TRIANGLE_CLASS:
-		m_topology = GL_TRIANGLES;
+		t = GL_TRIANGLES;
 		break;
 	default:
 		__assume(0);
 	}
 
+	dev->IASetPrimitiveTopology(t);
+}
+
+
+void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Source* tex)
+{
 	GSDrawingEnvironment& env = m_env;
 	GSDrawingContext* context = m_context;
 
@@ -103,7 +103,7 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 			GSVector4 s = GSVector4(rtscale.x / rtsize.x, rtscale.y / rtsize.y);
 			GSVector4 o = GSVector4(-1.0f, 1.0f);
 
-			GSVector4 src = ((m_vt->m_min.p.xyxy(m_vt->m_max.p) + o.xxyy()) * s.xyxy()).sat(o.zzyy());
+			GSVector4 src = ((m_vt.m_min.p.xyxy(m_vt.m_max.p) + o.xxyy()) * s.xyxy()).sat(o.zzyy());
 			GSVector4 dst = src * 2.0f + o.xxxx();
 
 			GSVertexPT1 vertices[] =
@@ -163,7 +163,7 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 
 	if(!IsOpaque())
 	{
-		om_bsel.abe = PRIM->ABE || PRIM->AA1 && m_vt->m_primclass == GS_LINE_CLASS;
+		om_bsel.abe = PRIM->ABE || PRIM->AA1 && m_vt.m_primclass == GS_LINE_CLASS;
 
 		om_bsel.a = context->ALPHA.A;
 		om_bsel.b = context->ALPHA.B;
@@ -207,11 +207,11 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 	{
 		if(context->ZBUF.PSM == PSM_PSMZ24)
 		{
-			if(m_vt->m_max.p.z > 0xffffff)
+			if(m_vt.m_max.p.z > 0xffffff)
 			{
-				ASSERT(m_vt->m_min.p.z > 0xffffff);
+				ASSERT(m_vt.m_min.p.z > 0xffffff);
 				// Fixme :Following conditional fixes some dialog frame in Wild Arms 3, but may not be what was intended.
-				if (m_vt->m_min.p.z > 0xffffff)
+				if (m_vt.m_min.p.z > 0xffffff)
 				{
 					vs_sel.bppz = 1;
 					om_dssel.ztst = ZTST_ALWAYS;
@@ -220,11 +220,11 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 		}
 		else if(context->ZBUF.PSM == PSM_PSMZ16 || context->ZBUF.PSM == PSM_PSMZ16S)
 		{
-			if(m_vt->m_max.p.z > 0xffff)
+			if(m_vt.m_max.p.z > 0xffff)
 			{
-				ASSERT(m_vt->m_min.p.z > 0xffff); // sfex capcom logo
+				ASSERT(m_vt.m_min.p.z > 0xffff); // sfex capcom logo
 				// Fixme : Same as above, I guess.
-				if (m_vt->m_min.p.z > 0xffff)
+				if (m_vt.m_min.p.z > 0xffff)
 				{
 					vs_sel.bppz = 2;
 					om_dssel.ztst = ZTST_ALWAYS;
@@ -268,7 +268,7 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 	GSDeviceOGL::GSSelector gs_sel;
 
 	gs_sel.iip = PRIM->IIP;
-	gs_sel.prim = m_vt->m_primclass;
+	gs_sel.prim = m_vt.m_primclass;
 
 	// ps
 
@@ -288,7 +288,7 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 		}
 	}
 
-	if (env.COLCLAMP.CLAMP == 0 && /* hack */ !tex && PRIM->PRIM != GS_POINTLIST)
+	if(env.COLCLAMP.CLAMP == 0 && /* hack */ !tex && PRIM->PRIM != GS_POINTLIST)
 	{
 		ps_sel.colclip = 1;
 	}
@@ -336,7 +336,7 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 		ps_sel.aem = env.TEXA.AEM;
 		ps_sel.tfx = context->TEX0.TFX;
 		ps_sel.tcc = context->TEX0.TCC;
-		ps_sel.ltf = m_filter == 2 ? m_vt->IsLinear() : m_filter;
+		ps_sel.ltf = m_filter == 2 ? m_vt.IsLinear() : m_filter;
 		ps_sel.rt = tex->m_target;
 
 		int w = tex->m_texture->GetWidth();
@@ -385,8 +385,9 @@ void GSRendererOGL::DrawPrims(GSTexture* rt, GSTexture* ds, GSTextureCache::Sour
 
 	uint8 afix = context->ALPHA.FIX;
 
+	SetupIA();
+
 	dev->SetupOM(om_dssel, om_bsel, afix);
-	dev->SetupIA(m_vertex.buff, m_vertex.next, m_index.buff, m_index.tail, m_topology);
 	dev->SetupVS(vs_sel, &vs_cb);
 	dev->SetupGS(gs_sel);
 	dev->SetupPS(ps_sel, &ps_cb, ps_ssel);
diff --git a/plugins/GSdx/GSRendererOGL.h b/plugins/GSdx/GSRendererOGL.h
index 62736d5c04..7cef97af87 100644
--- a/plugins/GSdx/GSRendererOGL.h
+++ b/plugins/GSdx/GSRendererOGL.h
@@ -40,22 +40,12 @@ class GSRendererOGL : public GSRendererHW
 		bool UserHacks_AlphaHack;
 
 	protected:
-		GLenum m_topology;
-
-		template<uint32 prim, uint32 tme, uint32 fst>
-		void ConvertVertex(size_t dst_index, size_t src_index);
-
-		int GetPosX(const void* vertex) const {return (int)((const GSVertexHW11*)vertex)->p.x;}
-		int GetPosY(const void* vertex) const {return (int)((const GSVertexHW11*)vertex)->p.y;}
-		uint32 GetColor(const void* vertex) const {return ((const GSVertexHW11*)vertex)->c0;}
-		void SetColor(void* vertex, uint32 c) const {((GSVertexHW11*)vertex)->c0 = c;}
+		void SetupIA();
 
 	public:
 		GSRendererOGL();
 		virtual ~GSRendererOGL() {};
 
-		template<uint32 prim, uint32 tme, uint32 fst> void VertexKick(bool skip);
-
 		bool CreateDevice(GSDevice* dev);
 
 		void UpdateFBA(GSTexture* rt) {}
diff --git a/plugins/GSdx/GSTextureCacheOGL.cpp b/plugins/GSdx/GSTextureCacheOGL.cpp
index b248d1dcf1..74eba438a0 100644
--- a/plugins/GSdx/GSTextureCacheOGL.cpp
+++ b/plugins/GSdx/GSTextureCacheOGL.cpp
@@ -62,6 +62,7 @@ void GSTextureCacheOGL::Read(Target* t, const GSVector4i& r)
 	GSVector4 src = GSVector4(r) * GSVector4(t->m_texture->GetScale()).xyxy() / GSVector4(t->m_texture->GetSize()).xyxy();
 
 	GLuint format = TEX0.PSM == PSM_PSMCT16 || TEX0.PSM == PSM_PSMCT16S ? GL_R16UI : GL_RGBA8;
+	//if (format == GL_R16UI) fprintf(stderr, "Format 16 bits integer\n");
 #if 0
 	DXGI_FORMAT format = TEX0.PSM == PSM_PSMCT16 || TEX0.PSM == PSM_PSMCT16S ? DXGI_FORMAT_R16_UINT : DXGI_FORMAT_R8G8B8A8_UNORM;
 #endif
diff --git a/plugins/GSdx/GSTextureCacheOGL.h b/plugins/GSdx/GSTextureCacheOGL.h
index 11bf6210ad..f86554c84a 100644
--- a/plugins/GSdx/GSTextureCacheOGL.h
+++ b/plugins/GSdx/GSTextureCacheOGL.h
@@ -28,7 +28,7 @@
 class GSTextureCacheOGL : public GSTextureCache
 {
 protected:
-	int Get8bitFormat() {/* TODO return DXGI_FORMAT_A8_UNORM;*/}
+	int Get8bitFormat() { return GL_R8; /* TODO return DXGI_FORMAT_A8_UNORM;*/}
 
 	void Read(Target* t, const GSVector4i& r);
 
diff --git a/plugins/GSdx/GSTextureFXOGL.cpp b/plugins/GSdx/GSTextureFXOGL.cpp
index 5616a72078..ba194377db 100644
--- a/plugins/GSdx/GSTextureFXOGL.cpp
+++ b/plugins/GSdx/GSTextureFXOGL.cpp
@@ -61,26 +61,19 @@ void GSDeviceOGL::CreateTextureFX()
 	//float4 c : COLOR0;
 	//float4 f : COLOR1;
 
-	GSInputLayoutOGL vert_format[6] =
+	GSInputLayoutOGL vert_format[] =
 	{
 		// FIXME
-		{0 , 2 , GL_FLOAT          , GL_FALSE , sizeof(GSVertexHW11) , (const GLvoid*)(0) }  ,
-		{1 , 4 , GL_UNSIGNED_BYTE  , GL_TRUE  , sizeof(GSVertexHW11) , (const GLvoid*)(8) }  ,
-		{2 , 1 , GL_FLOAT          , GL_FALSE , sizeof(GSVertexHW11) , (const GLvoid*)(12) } ,
-		{3 , 2 , GL_UNSIGNED_SHORT , GL_FALSE , sizeof(GSVertexHW11) , (const GLvoid*)(16) } ,
-		{4 , 1 , GL_UNSIGNED_INT   , GL_FALSE , sizeof(GSVertexHW11) , (const GLvoid*)(20) } ,
+		{0 , 2 , GL_FLOAT          , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(0) }  ,
+		{1 , 4 , GL_UNSIGNED_BYTE  , GL_TRUE  , sizeof(GSVertex) , (const GLvoid*)(8) }  ,
+		{2 , 1 , GL_FLOAT          , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(12) } ,
+		{3 , 2 , GL_UNSIGNED_SHORT , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(16) } ,
+		{4 , 1 , GL_UNSIGNED_INT   , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(20) } ,
 		// note: there is a 32 bits pad
-		{5 , 4 , GL_UNSIGNED_BYTE  , GL_TRUE  , sizeof(GSVertexHW11) , (const GLvoid*)(28) } ,
+		{5 , 2 , GL_UNSIGNED_SHORT , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(24) } ,
+		{6 , 4 , GL_UNSIGNED_BYTE  , GL_TRUE  , sizeof(GSVertex) , (const GLvoid*)(28) } ,
 	};
-	m_vb = new GSVertexBufferStateOGL(sizeof(GSVertexHW11), vert_format, countof(vert_format));
-}
-
-void GSDeviceOGL::SetupIA(const void* vertex, int vertex_count, const uint32* index, int index_count, int prim)
-{
-	IASetVertexState(m_vb);
-	IASetVertexBuffer(vertex, vertex_count);
-	IASetIndexBuffer(index, index_count);
-	IASetPrimitiveTopology(prim);
+	m_vb = new GSVertexBufferStateOGL(sizeof(GSVertex), vert_format, countof(vert_format));
 }
 
 void GSDeviceOGL::SetupVS(VSSelector sel, const VSConstantBuffer* cb)
diff --git a/plugins/GSdx/GSTextureOGL.cpp b/plugins/GSdx/GSTextureOGL.cpp
index d26c1d3f0d..2590bbc3be 100644
--- a/plugins/GSdx/GSTextureOGL.cpp
+++ b/plugins/GSdx/GSTextureOGL.cpp
@@ -103,11 +103,16 @@ GSTextureOGL::GSTextureOGL(int type, int w, int h, bool msaa, int format, GLuint
 			// In worst case the HW renderer seems to use 3 texture unit
 			// For the moment SW renderer only use 1 so don't bother
 			EnableUnit(2);
-			if (m_format == GL_RGBA8) {
+			if (m_format == GL_RGBA8)
 				glTexImage2D(m_texture_target, 0, m_format, m_size.x, m_size.y, 0, GL_RGBA, GL_UNSIGNED_BYTE, NULL);
-			}
-			else
+			else if (m_format == GL_R16UI)
+				glTexImage2D(m_texture_target, 0, m_format, m_size.x, m_size.y, 0, GL_RED_INTEGER, GL_UNSIGNED_SHORT, NULL);
+			else if (m_format == GL_R8)
+				glTexImage2D(m_texture_target, 0, m_format, m_size.x, m_size.y, 0, GL_RED, GL_UNSIGNED_BYTE, NULL);
+			else {
+				fprintf(stderr, "wrong texture pixel format :%x\n", m_format);
 				assert(0); // TODO Later
+			}
 			break;
 		case GSTexture::Offscreen:
 			if (m_type == GL_RGBA8) m_pbo_size = m_size.x * m_size.y * 4;
@@ -130,7 +135,10 @@ GSTextureOGL::~GSTextureOGL()
 
 void GSTextureOGL::Attach(GLenum attachment)
 {
+	//fprintf(stderr, "format %d,%x\n", m_type, m_format);
 	glFramebufferTexture2D(GL_FRAMEBUFFER, attachment, m_texture_target, m_texture_id, 0);
+	// FIXME DEBUG
+	//fprintf(stderr, "FB status %x\n", glCheckFramebufferStatus(GL_FRAMEBUFFER));
 }
 
 bool GSTextureOGL::Update(const GSVector4i& r, const void* data, int pitch)
@@ -142,18 +150,22 @@ bool GSTextureOGL::Update(const GSVector4i& r, const void* data, int pitch)
 
 	EnableUnit(2);
 
-	if (m_format != GL_RGBA8) {
-		fprintf(stderr, "wrong pixel format\n");
-		assert(0);
-	}
-
 	// pitch could be different of width*element_size
 	glPixelStorei(GL_UNPACK_ROW_LENGTH, pitch>>2);
 	// FIXME : it crash on colin mcrae rally 3 (others game too) when the size is 16
 	//fprintf(stderr, "Texture %dx%d with a pitch of %d\n", m_size.x, m_size.y, pitch >>2);
 	//fprintf(stderr, "Box (%d,%d)x(%d,%d)\n", r.x, r.y, r.width(), r.height());
 
-	glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RGBA, GL_UNSIGNED_BYTE, data);
+	if (m_format == GL_RGBA8)
+		glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RGBA, GL_UNSIGNED_BYTE, data);
+	else if (m_format == GL_R16UI)
+		glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RED_INTEGER, GL_R16UI, data);
+	else if (m_format == GL_R8)
+		glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RED, GL_R8, data);
+	else {
+		fprintf(stderr, "wrong texture pixel format :%x\n", m_format);
+		assert(0);
+	}
 #if 0
 	//if (m_size.x != 16)
 	if (r.width() > 16 && r.height() > 16)
@@ -216,12 +228,16 @@ bool GSTextureOGL::Map(GSMap& m, const GSVector4i* r)
 		// FIXME It might be possible to only map a subrange of the texture based on r object
 
 		// Load the PBO
-		if (m_format == GL_R16UI)
-			glReadPixels(0, 0, m_size.x, m_size.y, GL_RED, GL_UNSIGNED_SHORT, 0);
-		else if (m_format == GL_RGBA8)
+		if (m_format == GL_RGBA8)
 			glReadPixels(0, 0, m_size.x, m_size.y, GL_RGBA, GL_UNSIGNED_BYTE, 0);
-		else
+		else if (m_format == GL_R16UI)
+			glReadPixels(0, 0, m_size.x, m_size.y, GL_RED_INTEGER, GL_UNSIGNED_SHORT, 0);
+		else if (m_format == GL_R8)
+			glReadPixels(0, 0, m_size.x, m_size.y, GL_RED, GL_UNSIGNED_BYTE, 0);
+		else {
+			fprintf(stderr, "wrong texture pixel format :%x\n", m_format);
 			assert(0);
+		}
 
 		// Give access from the CPU
 		uint32 map_flags = GL_MAP_READ_BIT;
diff --git a/plugins/GSdx/GSThread.h b/plugins/GSdx/GSThread.h
index 4a9e7223dc..160bd541cd 100644
--- a/plugins/GSdx/GSThread.h
+++ b/plugins/GSdx/GSThread.h
@@ -182,6 +182,8 @@ public:
     void Lock() {pthread_mutex_lock(&m_mutex);}
     bool TryLock() {return pthread_mutex_trylock(&m_mutex) == 0;}
     void Unlock() {pthread_mutex_unlock(&m_mutex);}
+
+	operator pthread_mutex_t* () {return &m_mutex;}
 };
 
 class GSEvent : public IGSEvent
diff --git a/plugins/GSdx/res/tfx.glsl b/plugins/GSdx/res/tfx.glsl
index 892af3178e..c45dcd9063 100644
--- a/plugins/GSdx/res/tfx.glsl
+++ b/plugins/GSdx/res/tfx.glsl
@@ -48,12 +48,13 @@ struct vertex
 };
 
 #ifdef VERTEX_SHADER
-layout(location = 0) in vec2  i_t;
+layout(location = 0) in vec2  i_st;
 layout(location = 1) in vec4  i_c;
 layout(location = 2) in float i_q;
 layout(location = 3) in uvec2 i_p;
 layout(location = 4) in uint  i_z;
-layout(location = 5) in vec4  i_f;
+layout(location = 5) in uvec2 i_uv;
+layout(location = 6) in vec4  i_f;
 
 layout(location = 0) out vertex VSout;
 
@@ -101,12 +102,14 @@ void vs_main()
 	{
 		if(VS_FST != 0)
 		{
-			VSout.t.xy = i_t * TextureScale;
+			//VSout.t.xy = i_t * TextureScale;
+			VSout.t.xy = i_uv * TextureScale;
 			VSout.t.w = 1.0f;
 		}
 		else
 		{
-			VSout.t.xy = i_t;
+			//VSout.t.xy = i_t;
+			VSout.t.xy = i_st;
 			VSout.t.w = i_q;
 		}
 	}