Automated merge with file:///Users/nat/linden/viewer-maint-1144

21 files changed, 3109 insertions, 1318 deletions

diff --git a/indra/llrender/llcubemap.cpp b/indra/llrender/llcubemap.cpp
index 1b10354c22..45a3b18179 100644
--- a/indra/llrender/llcubemap.cpp
+++ b/indra/llrender/llcubemap.cpp
@@ -36,6 +36,7 @@
 #include "m4math.h"
 
 #include "llrender.h"
+#include "llglslshader.h"
 
 #include "llglheaders.h"
 
@@ -195,7 +196,7 @@ void LLCubeMap::enableTexture(S32 stage)
 void LLCubeMap::enableTextureCoords(S32 stage)
 {
 	mTextureCoordStage = stage;
-	if (gGLManager.mHasCubeMap && stage >= 0 && LLCubeMap::sUseCubeMaps)
+	if (!LLGLSLShader::sNoFixedFunction && gGLManager.mHasCubeMap && stage >= 0 && LLCubeMap::sUseCubeMaps)
 	{
 		if (stage > 0)
 		{
@@ -237,7 +238,7 @@ void LLCubeMap::disableTexture(void)
 
 void LLCubeMap::disableTextureCoords(void)
 {
-	if (gGLManager.mHasCubeMap && mTextureCoordStage >= 0 && LLCubeMap::sUseCubeMaps)
+	if (!LLGLSLShader::sNoFixedFunction && gGLManager.mHasCubeMap && mTextureCoordStage >= 0 && LLCubeMap::sUseCubeMaps)
 	{
 		if (mTextureCoordStage > 0)
 		{
@@ -264,19 +265,19 @@ void LLCubeMap::setMatrix(S32 stage)
 		gGL.getTexUnit(stage)->activate();
 	}
 
-	LLVector3 x(LLVector3d(gGLModelView+0));
-	LLVector3 y(LLVector3d(gGLModelView+4));
-	LLVector3 z(LLVector3d(gGLModelView+8));
+	LLVector3 x(gGLModelView+0);
+	LLVector3 y(gGLModelView+4);
+	LLVector3 z(gGLModelView+8);
 
 	LLMatrix3 mat3;
 	mat3.setRows(x,y,z);
 	LLMatrix4 trans(mat3);
 	trans.transpose();
 
-	glMatrixMode(GL_TEXTURE);
-	glPushMatrix();
-	glLoadMatrixf((F32 *)trans.mMatrix);
-	glMatrixMode(GL_MODELVIEW);
+	gGL.matrixMode(LLRender::MM_TEXTURE);
+	gGL.pushMatrix();
+	gGL.loadMatrix((F32 *)trans.mMatrix);
+	gGL.matrixMode(LLRender::MM_MODELVIEW);
 	
 	/*if (stage > 0)
 	{
@@ -292,9 +293,9 @@ void LLCubeMap::restoreMatrix()
 	{
 		gGL.getTexUnit(mMatrixStage)->activate();
 	}
-	glMatrixMode(GL_TEXTURE);
-	glPopMatrix();
-	glMatrixMode(GL_MODELVIEW);
+	gGL.matrixMode(LLRender::MM_TEXTURE);
+	gGL.popMatrix();
+	gGL.matrixMode(LLRender::MM_MODELVIEW);
 	
 	/*if (mMatrixStage > 0)
 	{
diff --git a/indra/llrender/llfontfreetype.cpp b/indra/llrender/llfontfreetype.cpp
index 91c8a37022..66d4ad2d87 100644
--- a/indra/llrender/llfontfreetype.cpp
+++ b/indra/llrender/llfontfreetype.cpp
@@ -55,7 +55,10 @@ FT_Library gFTLibrary = NULL;
 //static
 void LLFontManager::initClass()
 {
-	gFontManagerp = new LLFontManager;
+	if (!gFontManagerp) 
+	{
+		gFontManagerp = new LLFontManager;
+	}
 }
 
 //static
@@ -136,7 +139,7 @@ BOOL LLFontFreetype::loadFace(const std::string& filename, F32 point_size, F32 v
 		FT_Done_Face(mFTFace);
 		mFTFace = NULL;
 	}
-
+	
 	int error;
 
 	error = FT_New_Face( gFTLibrary,
diff --git a/indra/llrender/llfontgl.cpp b/indra/llrender/llfontgl.cpp
index 607473d416..fccbf37a8d 100644
--- a/indra/llrender/llfontgl.cpp
+++ b/indra/llrender/llfontgl.cpp
@@ -56,8 +56,9 @@ std::string LLFontGL::sAppDir;
 LLColor4 LLFontGL::sShadowColor(0.f, 0.f, 0.f, 1.f);
 LLFontRegistry* LLFontGL::sFontRegistry = NULL;
 
-LLCoordFont LLFontGL::sCurOrigin;
-std::vector<LLCoordFont> LLFontGL::sOriginStack;
+LLCoordGL LLFontGL::sCurOrigin;
+F32 LLFontGL::sCurDepth;
+std::vector<std::pair<LLCoordGL, F32> > LLFontGL::sOriginStack;
 
 const F32 EXT_X_BEARING = 1.f;
 const F32 EXT_Y_BEARING = 0.f;
@@ -68,20 +69,6 @@ const F32 PIXEL_CORRECTION_DISTANCE = 0.01f;
 const F32 PAD_UVY = 0.5f; // half of vertical padding between glyphs in the glyph texture
 const F32 DROP_SHADOW_SOFT_STRENGTH = 0.3f;
 
-static F32 llfont_round_x(F32 x)
-{
-	//return llfloor((x-LLFontGL::sCurOrigin.mX)/LLFontGL::sScaleX+0.5f)*LLFontGL::sScaleX+LLFontGL::sCurOrigin.mX;
-	//return llfloor(x/LLFontGL::sScaleX+0.5f)*LLFontGL::sScaleY;
-	return x;
-}
-
-static F32 llfont_round_y(F32 y)
-{
-	//return llfloor((y-LLFontGL::sCurOrigin.mY)/LLFontGL::sScaleY+0.5f)*LLFontGL::sScaleY+LLFontGL::sCurOrigin.mY;
-	//return llfloor(y+0.5f);
-	return y;
-}
-
 LLFontGL::LLFontGL()
 {
 }
@@ -115,23 +102,23 @@ static LLFastTimer::DeclareTimer FTM_RENDER_FONTS("Fonts");
 S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, const LLRect& rect, const LLColor4 &color, HAlign halign, VAlign valign, U8 style, 
 					 ShadowType shadow, S32 max_chars, F32* right_x, BOOL use_ellipses) const
 {
-	F32 x = rect.mLeft;
+	F32 x = (F32)rect.mLeft;
 	F32 y = 0.f;
 
 	switch(valign)
 	{
 	case TOP:
-		y = rect.mTop;
+		y = (F32)rect.mTop;
 		break;
 	case VCENTER:
-		y = rect.getCenterY();
+		y = (F32)rect.getCenterY();
 		break;
 	case BASELINE:
 	case BOTTOM:
-		y = rect.mBottom;
+		y = (F32)rect.mBottom;
 		break;
 	default:
-		y = rect.mBottom;
+		y = (F32)rect.mBottom;
 		break;
 	}
 	return render(wstr, begin_offset, x, y, color, halign, valign, style, shadow, max_chars, rect.getWidth(), right_x, use_ellipses);
@@ -177,21 +164,11 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 
 	gGL.loadUIIdentity();
 	
-	//gGL.translateUI(floorf(sCurOrigin.mX*sScaleX), floorf(sCurOrigin.mY*sScaleY), sCurOrigin.mZ);
-
-	// this code snaps the text origin to a pixel grid to start with
-	//F32 pixel_offset_x = llround((F32)sCurOrigin.mX) - (sCurOrigin.mX);
-	//F32 pixel_offset_y = llround((F32)sCurOrigin.mY) - (sCurOrigin.mY);
-	//gGL.translateUI(-pixel_offset_x, -pixel_offset_y, 0.f);
-
 	LLVector2 origin(floorf(sCurOrigin.mX*sScaleX), floorf(sCurOrigin.mY*sScaleY));
-	// snap the text origin to a pixel grid to start with
-	origin.mV[VX] -= llround((F32)sCurOrigin.mX) - (sCurOrigin.mX);
-	origin.mV[VY] -= llround((F32)sCurOrigin.mY) - (sCurOrigin.mY);
 
-	// Depth translation, so that floating text appears 'inworld'
-	// and is correclty occluded.
-	gGL.translatef(0.f,0.f,sCurOrigin.mZ);
+	// Depth translation, so that floating text appears 'in-world'
+	// and is correctly occluded.
+	gGL.translatef(0.f,0.f,sCurDepth);
 
 	S32 chars_drawn = 0;
 	S32 i;
@@ -215,16 +192,17 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 	cur_y = ((F32)y * sScaleY) + origin.mV[VY];
 
 	// Offset y by vertical alignment.
+	// use unscaled font metrics here
 	switch (valign)
 	{
 	case TOP:
-		cur_y -= mFontFreetype->getAscenderHeight();
+		cur_y -= llceil(mFontFreetype->getAscenderHeight());
 		break;
 	case BOTTOM:
-		cur_y += mFontFreetype->getDescenderHeight();
+		cur_y += llceil(mFontFreetype->getDescenderHeight());
 		break;
 	case VCENTER:
-		cur_y -= (mFontFreetype->getAscenderHeight() - mFontFreetype->getDescenderHeight()) / 2.f;
+		cur_y -= llceil((llceil(mFontFreetype->getAscenderHeight()) - llceil(mFontFreetype->getDescenderHeight())) / 2.f);
 		break;
 	case BASELINE:
 		// Baseline, do nothing.
@@ -250,7 +228,7 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 	cur_render_y = cur_y;
 	cur_render_x = cur_x;
 
-	F32 start_x = llround(cur_x);
+	F32 start_x = (F32)llround(cur_x);
 
 	const LLFontBitmapCache* font_bitmap_cache = mFontFreetype->getFontBitmapCache();
 
@@ -304,6 +282,18 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 		S32 next_bitmap_num = fgi->mBitmapNum;
 		if (next_bitmap_num != bitmap_num)
 		{
+			// Actually draw the queued glyphs before switching their texture;
+			// otherwise the queued glyphs will be taken from wrong textures.
+			if (glyph_count > 0)
+			{
+				gGL.begin(LLRender::QUADS);
+				{
+					gGL.vertexBatchPreTransformed(vertices, uvs, colors, glyph_count * 4);
+				}
+				gGL.end();
+				glyph_count = 0;
+			}
+
 			bitmap_num = next_bitmap_num;
 			LLImageGL *font_image = font_bitmap_cache->getImageGL(bitmap_num);
 			gGL.getTexUnit(0)->bind(font_image);
@@ -322,10 +312,10 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 				(fgi->mXBitmapOffset + fgi->mWidth) * inv_width,
 				(fgi->mYBitmapOffset - PAD_UVY) * inv_height);
 		// snap glyph origin to whole screen pixel
-		LLRectf screen_rect(llround(cur_render_x + (F32)fgi->mXBearing),
-				    llround(cur_render_y + (F32)fgi->mYBearing),
-				    llround(cur_render_x + (F32)fgi->mXBearing) + (F32)fgi->mWidth,
-				    llround(cur_render_y + (F32)fgi->mYBearing) - (F32)fgi->mHeight);
+		LLRectf screen_rect((F32)llround(cur_render_x + (F32)fgi->mXBearing),
+				    (F32)llround(cur_render_y + (F32)fgi->mYBearing),
+				    (F32)llround(cur_render_x + (F32)fgi->mXBearing) + (F32)fgi->mWidth,
+				    (F32)llround(cur_render_y + (F32)fgi->mYBearing) - (F32)fgi->mHeight);
 		
 		if (glyph_count >= GLYPH_BATCH_SIZE)
 		{
@@ -378,12 +368,12 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 	//FIXME: add underline as glyph?
 	if (style_to_add & UNDERLINE)
 	{
-		F32 descender = mFontFreetype->getDescenderHeight();
+		F32 descender = (F32)llfloor(mFontFreetype->getDescenderHeight());
 
 		gGL.getTexUnit(0)->unbind(LLTexUnit::TT_TEXTURE);
 		gGL.begin(LLRender::LINES);
-		gGL.vertex2f(start_x, cur_y - (descender));
-		gGL.vertex2f(cur_x, cur_y - (descender));
+		gGL.vertex2f(start_x, cur_y - descender);
+		gGL.vertex2f(cur_x, cur_y - descender);
 		gGL.end();
 	}
 
@@ -432,19 +422,9 @@ S32 LLFontGL::renderUTF8(const std::string &text, S32 begin_offset, S32 x, S32 y
 }
 
 // font metrics - override for LLFontFreetype that returns units of virtual pixels
-F32 LLFontGL::getLineHeight() const
-{ 
-	return (F32)llround(mFontFreetype->getLineHeight() / sScaleY); 
-}
-
-F32 LLFontGL::getAscenderHeight() const
+S32 LLFontGL::getLineHeight() const
 { 
-	return (F32)llround(mFontFreetype->getAscenderHeight() / sScaleY); 
-}
-
-F32 LLFontGL::getDescenderHeight() const
-{ 
-	return (F32)llround(mFontFreetype->getDescenderHeight() / sScaleY); 
+	return llceil(mFontFreetype->getAscenderHeight() / sScaleY) + llceil(mFontFreetype->getDescenderHeight() / sScaleY);
 }
 
 S32 LLFontGL::getWidth(const std::string& utf8text) const
@@ -633,7 +613,7 @@ S32 LLFontGL::maxDrawableChars(const llwchar* wchars, F32 max_pixels, S32 max_ch
 		}
 
 		// Round after kerning.
-		cur_x = llround(cur_x);
+		cur_x = (F32)llround(cur_x);
 		drawn_x = cur_x;
 	}
 
@@ -704,7 +684,7 @@ S32	LLFontGL::firstDrawableChar(const llwchar* wchars, F32 max_pixels, S32 text_
 		}
 
 		// Round after kerning.
-		total_width = llround(total_width);
+		total_width = (F32)llround(total_width);
 	}
 
 	if (drawable_chars == 0)
@@ -787,7 +767,7 @@ S32 LLFontGL::charFromPixelOffset(const llwchar* wchars, S32 begin_offset, F32 t
 
 
 		// Round after kerning.
-		cur_x = llround(cur_x);
+		cur_x = (F32)llround(cur_x);
 	}
 
 	return llmin(max_chars, pos - begin_offset);
@@ -1134,22 +1114,22 @@ void LLFontGL::renderQuad(LLVector3* vertex_out, LLVector2* uv_out, LLColor4U* c
 {
 	S32 index = 0;
 
-	vertex_out[index] = LLVector3(llfont_round_x(screen_rect.mRight), llfont_round_y(screen_rect.mTop), 0.f);
+	vertex_out[index] = LLVector3(screen_rect.mRight, screen_rect.mTop, 0.f);
 	uv_out[index] = LLVector2(uv_rect.mRight, uv_rect.mTop);
 	colors_out[index] = color;
 	index++;
 
-	vertex_out[index] = LLVector3(llfont_round_x(screen_rect.mLeft), llfont_round_y(screen_rect.mTop), 0.f);
+	vertex_out[index] = LLVector3(screen_rect.mLeft, screen_rect.mTop, 0.f);
 	uv_out[index] = LLVector2(uv_rect.mLeft, uv_rect.mTop);
 	colors_out[index] = color;
 	index++;
 
-	vertex_out[index] = LLVector3(llfont_round_x(screen_rect.mLeft), llfont_round_y(screen_rect.mBottom), 0.f);
+	vertex_out[index] = LLVector3(screen_rect.mLeft, screen_rect.mBottom, 0.f);
 	uv_out[index] = LLVector2(uv_rect.mLeft, uv_rect.mBottom);
 	colors_out[index] = color;
 	index++;
 
-	vertex_out[index] = LLVector3(llfont_round_x(screen_rect.mRight), llfont_round_y(screen_rect.mBottom), 0.f);
+	vertex_out[index] = LLVector3(screen_rect.mRight, screen_rect.mBottom, 0.f);
 	uv_out[index] = LLVector2(uv_rect.mRight, uv_rect.mBottom);
 	colors_out[index] = color;
 }
diff --git a/indra/llrender/llfontgl.h b/indra/llrender/llfontgl.h
index dc8d848ed2..74bdbb43e7 100644
--- a/indra/llrender/llfontgl.h
+++ b/indra/llrender/llfontgl.h
@@ -115,9 +115,7 @@ public:
 	S32 renderUTF8(const std::string &text, S32 begin_offset, S32 x, S32 y, const LLColor4 &color, HAlign halign, VAlign valign, U8 style = NORMAL, ShadowType shadow = NO_SHADOW) const;
 
 	// font metrics - override for LLFontFreetype that returns units of virtual pixels
-	F32 getLineHeight() const;
-	F32 getAscenderHeight() const;
-	F32 getDescenderHeight() const;
+	S32 getLineHeight() const;
 
 	S32 getWidth(const std::string& utf8text) const;
 	S32 getWidth(const llwchar* wchars) const;
@@ -188,8 +186,9 @@ public:
 	static std::string getFontPathLocal();
 	static std::string getFontPathSystem();
 
-	static LLCoordFont sCurOrigin;
-	static std::vector<LLCoordFont> sOriginStack;
+	static LLCoordGL sCurOrigin;
+	static F32			sCurDepth;
+	static std::vector<std::pair<LLCoordGL, F32> > sOriginStack;
 
 	static LLColor4 sShadowColor;
 
diff --git a/indra/llrender/llgl.cpp b/indra/llrender/llgl.cpp
index 4e3cfb9c8a..197bc2b422 100644
--- a/indra/llrender/llgl.cpp
+++ b/indra/llrender/llgl.cpp
@@ -67,6 +67,38 @@ static const std::string HEADLESS_VERSION_STRING("1.0");
 
 std::ofstream gFailLog;
 
+#if GL_ARB_debug_output
+
+#ifndef APIENTRY
+#define APIENTRY
+#endif
+
+void APIENTRY gl_debug_callback(GLenum source,
+                                GLenum type,
+                                GLuint id,
+                                GLenum severity,
+                                GLsizei length,
+                                const GLchar* message,
+                                GLvoid* userParam)
+{
+	if (severity == GL_DEBUG_SEVERITY_HIGH_ARB)
+	{
+		llwarns << "----- GL ERROR --------" << llendl;
+	}
+	else
+	{
+		llwarns << "----- GL WARNING -------" << llendl;
+	}
+	llwarns << "Type: " << std::hex << type << llendl;
+	llwarns << "ID: " << std::hex << id << llendl;
+	llwarns << "Severity: " << std::hex << severity << llendl;
+	llwarns << "Message: " << message << llendl;
+	llwarns << "-----------------------" << llendl;
+}
+#endif
+
+void parse_glsl_version(S32& major, S32& minor);
+
 void ll_init_fail_log(std::string filename)
 {
 	gFailLog.open(filename.c_str());
@@ -110,6 +142,11 @@ std::list<LLGLUpdate*> LLGLUpdate::sGLQ;
 
 #if (LL_WINDOWS || LL_LINUX || LL_SOLARIS)  && !LL_MESA_HEADLESS
 // ATI prototypes
+
+#if LL_WINDOWS
+PFNGLGETSTRINGIPROC glGetStringi = NULL;
+#endif
+
 // vertex blending prototypes
 PFNGLWEIGHTPOINTERARBPROC			glWeightPointerARB = NULL;
 PFNGLVERTEXBLENDARBPROC				glVertexBlendARB = NULL;
@@ -128,6 +165,12 @@ PFNGLUNMAPBUFFERARBPROC				glUnmapBufferARB = NULL;
 PFNGLGETBUFFERPARAMETERIVARBPROC	glGetBufferParameterivARB = NULL;
 PFNGLGETBUFFERPOINTERVARBPROC		glGetBufferPointervARB = NULL;
 
+//GL_ARB_vertex_array_object
+PFNGLBINDVERTEXARRAYPROC glBindVertexArray = NULL;
+PFNGLDELETEVERTEXARRAYSPROC glDeleteVertexArrays = NULL;
+PFNGLGENVERTEXARRAYSPROC glGenVertexArrays = NULL;
+PFNGLISVERTEXARRAYPROC glIsVertexArray = NULL;
+
 // GL_ARB_map_buffer_range
 PFNGLMAPBUFFERRANGEPROC			glMapBufferRange = NULL;
 PFNGLFLUSHMAPPEDBUFFERRANGEPROC	glFlushMappedBufferRange = NULL;
@@ -197,10 +240,16 @@ PFNGLRENDERBUFFERSTORAGEMULTISAMPLEPROC glRenderbufferStorageMultisample = NULL;
 PFNGLFRAMEBUFFERTEXTURELAYERPROC glFramebufferTextureLayer = NULL;
 
 //GL_ARB_texture_multisample
-PFNGLTEXIMAGE2DMULTISAMPLEPROC glTexImage2DMultisample;
-PFNGLTEXIMAGE3DMULTISAMPLEPROC glTexImage3DMultisample;
-PFNGLGETMULTISAMPLEFVPROC glGetMultisamplefv;
-PFNGLSAMPLEMASKIPROC glSampleMaski;
+PFNGLTEXIMAGE2DMULTISAMPLEPROC glTexImage2DMultisample = NULL;
+PFNGLTEXIMAGE3DMULTISAMPLEPROC glTexImage3DMultisample = NULL;
+PFNGLGETMULTISAMPLEFVPROC glGetMultisamplefv = NULL;
+PFNGLSAMPLEMASKIPROC glSampleMaski = NULL;
+
+//GL_ARB_debug_output
+PFNGLDEBUGMESSAGECONTROLARBPROC glDebugMessageControlARB = NULL;
+PFNGLDEBUGMESSAGEINSERTARBPROC glDebugMessageInsertARB = NULL;
+PFNGLDEBUGMESSAGECALLBACKARBPROC glDebugMessageCallbackARB = NULL;
+PFNGLGETDEBUGMESSAGELOGARBPROC glGetDebugMessageLogARB = NULL;
 
 // GL_EXT_blend_func_separate
 PFNGLBLENDFUNCSEPARATEEXTPROC glBlendFuncSeparateEXT = NULL;
@@ -248,6 +297,11 @@ PFNGLGETACTIVEUNIFORMARBPROC glGetActiveUniformARB = NULL;
 PFNGLGETUNIFORMFVARBPROC glGetUniformfvARB = NULL;
 PFNGLGETUNIFORMIVARBPROC glGetUniformivARB = NULL;
 PFNGLGETSHADERSOURCEARBPROC glGetShaderSourceARB = NULL;
+PFNGLVERTEXATTRIBIPOINTERPROC glVertexAttribIPointer = NULL;
+
+#if LL_WINDOWS
+PFNWGLCREATECONTEXTATTRIBSARBPROC wglCreateContextAttribsARB = NULL;
+#endif
 
 // vertex shader prototypes
 #if LL_LINUX || LL_SOLARIS
@@ -349,6 +403,7 @@ LLGLManager::LLGLManager() :
 	mHasBlendFuncSeparate(FALSE),
 	mHasSync(FALSE),
 	mHasVertexBufferObject(FALSE),
+	mHasVertexArrayObject(FALSE),
 	mHasMapBufferRange(FALSE),
 	mHasFlushBufferRange(FALSE),
 	mHasPBuffer(FALSE),
@@ -370,6 +425,7 @@ LLGLManager::LLGLManager() :
 	mHasAnisotropic(FALSE),
 	mHasARBEnvCombine(FALSE),
 	mHasCubeMap(FALSE),
+	mHasDebugOutput(FALSE),
 
 	mIsATI(FALSE),
 	mIsNVIDIA(FALSE),
@@ -390,7 +446,8 @@ LLGLManager::LLGLManager() :
 	mDriverVersionMinor(0),
 	mDriverVersionRelease(0),
 	mGLVersion(1.0f),
-		
+	mGLSLVersionMajor(0),
+	mGLSLVersionMinor(0),		
 	mVRAM(0),
 	mGLMaxVertexRange(0),
 	mGLMaxIndexRange(0)
@@ -409,6 +466,15 @@ void LLGLManager::initWGL()
 		LL_WARNS("RenderInit") << "No ARB pixel format extensions" << LL_ENDL;
 	}
 
+	if (ExtensionExists("WGL_ARB_create_context",gGLHExts.mSysExts))
+	{
+		GLH_EXT_NAME(wglCreateContextAttribsARB) = (PFNWGLCREATECONTEXTATTRIBSARBPROC)GLH_EXT_GET_PROC_ADDRESS("wglCreateContextAttribsARB");
+	}
+	else
+	{
+		LL_WARNS("RenderInit") << "No ARB create context extensions" << LL_ENDL;
+	}
+	
 	if (ExtensionExists("WGL_EXT_swap_control", gGLHExts.mSysExts))
 	{
         GLH_EXT_NAME(wglSwapIntervalEXT) = (PFNWGLSWAPINTERVALEXTPROC)GLH_EXT_GET_PROC_ADDRESS("wglSwapIntervalEXT");
@@ -438,13 +504,45 @@ bool LLGLManager::initGL()
 		LL_ERRS("RenderInit") << "Calling init on LLGLManager after already initialized!" << LL_ENDL;
 	}
 
-	GLint alpha_bits;
-	glGetIntegerv( GL_ALPHA_BITS, &alpha_bits );
-	if( 8 != alpha_bits )
+	stop_glerror();
+
+#if LL_WINDOWS
+	if (!glGetStringi)
 	{
-		LL_WARNS("RenderInit") << "Frame buffer has less than 8 bits of alpha.  Avatar texture compositing will fail." << LL_ENDL;
+		glGetStringi = (PFNGLGETSTRINGIPROC) GLH_EXT_GET_PROC_ADDRESS("glGetStringi");
 	}
 
+	//reload extensions string (may have changed after using wglCreateContextAttrib)
+	if (glGetStringi)
+	{
+		std::stringstream str;
+
+		GLint count = 0;
+		glGetIntegerv(GL_NUM_EXTENSIONS, &count);
+		for (GLint i = 0; i < count; ++i)
+		{
+			std::string ext((const char*) glGetStringi(GL_EXTENSIONS, i));
+			str << ext << " ";
+			LL_DEBUGS("GLExtensions") << ext << llendl;
+		}
+		
+		{
+			PFNWGLGETEXTENSIONSSTRINGARBPROC wglGetExtensionsStringARB = 0;
+			wglGetExtensionsStringARB = (PFNWGLGETEXTENSIONSSTRINGARBPROC)wglGetProcAddress("wglGetExtensionsStringARB");
+			if(wglGetExtensionsStringARB)
+			{
+				str << (const char*) wglGetExtensionsStringARB(wglGetCurrentDC());
+			}
+		}
+
+		free(gGLHExts.mSysExts);
+		std::string extensions = str.str();
+		gGLHExts.mSysExts = strdup(extensions.c_str());
+	}
+#endif
+	
+	stop_glerror();
+
 	// Extract video card strings and convert to upper case to
 	// work around driver-to-driver variation in capitalization.
 	mGLVendor = std::string((const char *)glGetString(GL_VENDOR));
@@ -459,7 +557,21 @@ bool LLGLManager::initGL()
 		&mDriverVersionVendorString );
 
 	mGLVersion = mDriverVersionMajor + mDriverVersionMinor * .1f;
-	
+
+	if (mGLVersion >= 2.f)
+	{
+		parse_glsl_version(mGLSLVersionMajor, mGLSLVersionMinor);
+
+#if LL_DARWIN
+		//never use GLSL greater than 1.20 on OSX
+		if (mGLSLVersionMajor > 1 || mGLSLVersionMinor >= 30)
+		{
+			mGLSLVersionMajor = 1;
+			mGLSLVersionMinor = 20;
+		}
+#endif
+	}
+
 	// Trailing space necessary to keep "nVidia Corpor_ati_on" cards
 	// from being recognized as ATI.
 	if (mGLVendor.substr(0,4) == "ATI ")
@@ -531,8 +643,12 @@ bool LLGLManager::initGL()
 		mGLVendorShort = "MISC";
 	}
 	
+	stop_glerror();
 	// This is called here because it depends on the setting of mIsGF2or4MX, and sets up mHasMultitexture.
 	initExtensions();
+	stop_glerror();
+
+	S32 old_vram = mVRAM;
 
 	if (mHasATIMemInfo)
 	{ //ask the gl how much vram is free at startup and attempt to use no more than half of that
@@ -548,7 +664,27 @@ bool LLGLManager::initGL()
 		mVRAM = dedicated_memory/1024;
 	}
 
-	if (mHasMultitexture)
+	if (mVRAM < 256)
+	{ //something likely went wrong using the above extensions, fall back to old method
+		mVRAM = old_vram;
+	}
+
+	stop_glerror();
+
+	stop_glerror();
+
+	if (mHasFragmentShader)
+	{
+		GLint num_tex_image_units;
+		glGetIntegerv(GL_MAX_TEXTURE_IMAGE_UNITS_ARB, &num_tex_image_units);
+		mNumTextureImageUnits = llmin(num_tex_image_units, 32);
+	}
+
+	if (LLRender::sGLCoreProfile)
+	{
+		mNumTextureUnits = llmin(mNumTextureImageUnits, MAX_GL_TEXTURE_UNITS);
+	}
+	else if (mHasMultitexture)
 	{
 		GLint num_tex_units;		
 		glGetIntegerv(GL_MAX_TEXTURE_UNITS_ARB, &num_tex_units);
@@ -567,12 +703,7 @@ bool LLGLManager::initGL()
 		return false;
 	}
 	
-	if (mHasFragmentShader)
-	{
-		GLint num_tex_image_units;
-		glGetIntegerv(GL_MAX_TEXTURE_IMAGE_UNITS_ARB, &num_tex_image_units);
-		mNumTextureImageUnits = llmin(num_tex_image_units, 32);
-	}
+	stop_glerror();
 
 	if (mHasTextureMultisample)
 	{
@@ -582,6 +713,21 @@ bool LLGLManager::initGL()
 		glGetIntegerv(GL_MAX_SAMPLE_MASK_WORDS, &mMaxSampleMaskWords);
 	}
 
+	stop_glerror();
+
+#if LL_WINDOWS
+	if (mHasDebugOutput && gDebugGL)
+	{ //setup debug output callback
+		//glDebugMessageControlARB(GL_DONT_CARE, GL_DONT_CARE, GL_DEBUG_SEVERITY_LOW_ARB, 0, NULL, GL_TRUE);
+		glDebugMessageCallbackARB((GLDEBUGPROCARB) gl_debug_callback, NULL);
+		glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB);
+	}
+#endif
+
+	stop_glerror();
+
+	//HACK always disable texture multisample, use FXAA instead
+	mHasTextureMultisample = FALSE;
 #if LL_WINDOWS
 	if (mIsATI)
 	{ //using multisample textures on ATI results in black screen for some reason
@@ -593,10 +739,17 @@ bool LLGLManager::initGL()
 	{
 		glGetIntegerv(GL_MAX_SAMPLES, &mMaxSamples);
 	}
+
+	stop_glerror();
 	
 	setToDebugGPU();
 
+	stop_glerror();
+
 	initGLStates();
+
+	stop_glerror();
+
 	return true;
 }
 
@@ -700,14 +853,6 @@ std::string LLGLManager::getRawGLString()
 	return gl_string;
 }
 
-U32 LLGLManager::getNumFBOFSAASamples(U32 samples)
-{
-	samples = llmin(samples, (U32) mMaxColorTextureSamples);
-	samples = llmin(samples, (U32) mMaxDepthTextureSamples);
-	samples = llmin(samples, (U32) 4);
-	return samples;
-}
-
 void LLGLManager::shutdownGL()
 {
 	if (mInited)
@@ -774,7 +919,7 @@ void LLGLManager::initExtensions()
 	mHasVertexShader = FALSE;
 	mHasFragmentShader = FALSE;
 	mHasTextureRectangle = FALSE;
-#else // LL_MESA_HEADLESS
+#else // LL_MESA_HEADLESS //important, gGLHExts.mSysExts is uninitialized until after glh_init_extensions is called
 	mHasMultitexture = glh_init_extensions("GL_ARB_multitexture");
 	mHasATIMemInfo = ExtensionExists("GL_ATI_meminfo", gGLHExts.mSysExts);
 	mHasNVXMemInfo = ExtensionExists("GL_NVX_gpu_memory_info", gGLHExts.mSysExts);
@@ -788,6 +933,7 @@ void LLGLManager::initExtensions()
 	mHasOcclusionQuery = ExtensionExists("GL_ARB_occlusion_query", gGLHExts.mSysExts);
 	mHasOcclusionQuery2 = ExtensionExists("GL_ARB_occlusion_query2", gGLHExts.mSysExts);
 	mHasVertexBufferObject = ExtensionExists("GL_ARB_vertex_buffer_object", gGLHExts.mSysExts);
+	mHasVertexArrayObject = ExtensionExists("GL_ARB_vertex_array_object", gGLHExts.mSysExts);
 	mHasSync = ExtensionExists("GL_ARB_sync", gGLHExts.mSysExts);
 	mHasMapBufferRange = ExtensionExists("GL_ARB_map_buffer_range", gGLHExts.mSysExts);
 	mHasFlushBufferRange = ExtensionExists("GL_APPLE_flush_buffer_range", gGLHExts.mSysExts);
@@ -806,13 +952,14 @@ void LLGLManager::initExtensions()
 	mHasBlendFuncSeparate = ExtensionExists("GL_EXT_blend_func_separate", gGLHExts.mSysExts);
 	mHasTextureRectangle = ExtensionExists("GL_ARB_texture_rectangle", gGLHExts.mSysExts);
 	mHasTextureMultisample = ExtensionExists("GL_ARB_texture_multisample", gGLHExts.mSysExts);
+	mHasDebugOutput = ExtensionExists("GL_ARB_debug_output", gGLHExts.mSysExts);
 #if !LL_DARWIN
 	mHasPointParameters = !mIsATI && ExtensionExists("GL_ARB_point_parameters", gGLHExts.mSysExts);
 #endif
-	mHasShaderObjects = ExtensionExists("GL_ARB_shader_objects", gGLHExts.mSysExts) && ExtensionExists("GL_ARB_shading_language_100", gGLHExts.mSysExts);
+	mHasShaderObjects = ExtensionExists("GL_ARB_shader_objects", gGLHExts.mSysExts) && (LLRender::sGLCoreProfile || ExtensionExists("GL_ARB_shading_language_100", gGLHExts.mSysExts));
 	mHasVertexShader = ExtensionExists("GL_ARB_vertex_program", gGLHExts.mSysExts) && ExtensionExists("GL_ARB_vertex_shader", gGLHExts.mSysExts)
-						&& ExtensionExists("GL_ARB_shading_language_100", gGLHExts.mSysExts);
-	mHasFragmentShader = ExtensionExists("GL_ARB_fragment_shader", gGLHExts.mSysExts) && ExtensionExists("GL_ARB_shading_language_100", gGLHExts.mSysExts);
+		&& (LLRender::sGLCoreProfile || ExtensionExists("GL_ARB_shading_language_100", gGLHExts.mSysExts));
+	mHasFragmentShader = ExtensionExists("GL_ARB_fragment_shader", gGLHExts.mSysExts) && (LLRender::sGLCoreProfile || ExtensionExists("GL_ARB_shading_language_100", gGLHExts.mSysExts));
 #endif
 
 #if LL_LINUX || LL_SOLARIS
@@ -985,6 +1132,13 @@ void LLGLManager::initExtensions()
 			mHasVertexBufferObject = FALSE;
 		}
 	}
+	if (mHasVertexArrayObject)
+	{
+		glBindVertexArray = (PFNGLBINDVERTEXARRAYPROC) GLH_EXT_GET_PROC_ADDRESS("glBindVertexArray");
+		glDeleteVertexArrays = (PFNGLDELETEVERTEXARRAYSPROC) GLH_EXT_GET_PROC_ADDRESS("glDeleteVertexArrays");
+		glGenVertexArrays = (PFNGLGENVERTEXARRAYSPROC) GLH_EXT_GET_PROC_ADDRESS("glGenVertexArrays");
+		glIsVertexArray = (PFNGLISVERTEXARRAYPROC) GLH_EXT_GET_PROC_ADDRESS("glIsVertexArray");
+	}
 	if (mHasSync)
 	{
 		glFenceSync = (PFNGLFENCESYNCPROC) GLH_EXT_GET_PROC_ADDRESS("glFenceSync");
@@ -1039,6 +1193,13 @@ void LLGLManager::initExtensions()
 		glGetMultisamplefv = (PFNGLGETMULTISAMPLEFVPROC) GLH_EXT_GET_PROC_ADDRESS("glGetMultisamplefv");
 		glSampleMaski = (PFNGLSAMPLEMASKIPROC) GLH_EXT_GET_PROC_ADDRESS("glSampleMaski");
 	}	
+	if (mHasDebugOutput)
+	{
+		glDebugMessageControlARB = (PFNGLDEBUGMESSAGECONTROLARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDebugMessageControlARB");
+		glDebugMessageInsertARB = (PFNGLDEBUGMESSAGEINSERTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDebugMessageInsertARB");
+		glDebugMessageCallbackARB = (PFNGLDEBUGMESSAGECALLBACKARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDebugMessageCallbackARB");
+		glGetDebugMessageLogARB = (PFNGLGETDEBUGMESSAGELOGARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetDebugMessageLogARB");
+	}
 #if (!LL_LINUX && !LL_SOLARIS) || LL_LINUX_NV_GL_HEADERS
 	// This is expected to be a static symbol on Linux GL implementations, except if we use the nvidia headers - bah
 	glDrawRangeElements = (PFNGLDRAWRANGEELEMENTSPROC)GLH_EXT_GET_PROC_ADDRESS("glDrawRangeElements");
@@ -1157,6 +1318,7 @@ void LLGLManager::initExtensions()
 		glVertexAttrib4uivARB = (PFNGLVERTEXATTRIB4UIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4uivARB");
 		glVertexAttrib4usvARB = (PFNGLVERTEXATTRIB4USVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4usvARB");
 		glVertexAttribPointerARB = (PFNGLVERTEXATTRIBPOINTERARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttribPointerARB");
+		glVertexAttribIPointer = (PFNGLVERTEXATTRIBIPOINTERPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttribIPointer");
 		glEnableVertexAttribArrayARB = (PFNGLENABLEVERTEXATTRIBARRAYARBPROC) GLH_EXT_GET_PROC_ADDRESS("glEnableVertexAttribArrayARB");
 		glDisableVertexAttribArrayARB = (PFNGLDISABLEVERTEXATTRIBARRAYARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDisableVertexAttribArrayARB");
 		glProgramStringARB = (PFNGLPROGRAMSTRINGARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramStringARB");
@@ -1193,7 +1355,7 @@ void rotate_quat(LLQuaternion& rotation)
 {
 	F32 angle_radians, x, y, z;
 	rotation.getAngleAxis(&angle_radians, &x, &y, &z);
-	glRotatef(angle_radians * RAD_TO_DEG, x, y, z);
+	gGL.rotatef(angle_radians * RAD_TO_DEG, x, y, z);
 }
 
 void flush_glerror()
@@ -1230,10 +1392,6 @@ void log_glerror()
 
 void do_assert_glerror()
 {
-	if (LL_UNLIKELY(!gGLManager.mInited))
-	{
-		LL_ERRS("RenderInit") << "GL not initialized" << LL_ENDL;
-	}
 	//  Create or update texture to be used with this data 
 	GLenum error;
 	error = glGetError();
@@ -1326,11 +1484,6 @@ void LLGLState::initClass()
 	//make sure multisample defaults to disabled
 	sStateMap[GL_MULTISAMPLE_ARB] = GL_FALSE;
 	glDisable(GL_MULTISAMPLE_ARB);
-
-	sStateMap[GL_MULTISAMPLE_ARB] = GL_FALSE;
-	glDisable(GL_MULTISAMPLE_ARB);
-
-	glEnableClientState(GL_VERTEX_ARRAY);
 }
 
 //static
@@ -1604,7 +1757,7 @@ void LLGLState::checkTextureChannels(const std::string& msg)
 
 void LLGLState::checkClientArrays(const std::string& msg, U32 data_mask)
 {
-	if (!gDebugGL)
+	if (!gDebugGL || LLGLSLShader::sNoFixedFunction)
 	{
 		return;
 	}
@@ -1625,7 +1778,7 @@ void LLGLState::checkClientArrays(const std::string& msg, U32 data_mask)
 		error = TRUE;
 	}
 
-	glGetIntegerv(GL_ACTIVE_TEXTURE_ARB, &active_texture);
+	/*glGetIntegerv(GL_ACTIVE_TEXTURE_ARB, &active_texture);
 	if (active_texture != GL_TEXTURE0_ARB)
 	{
 		llwarns << "Active texture corrupted: " << active_texture << llendl;
@@ -1634,7 +1787,7 @@ void LLGLState::checkClientArrays(const std::string& msg, U32 data_mask)
 			gFailLog << "Active texture corrupted: " << active_texture << std::endl;
 		}
 		error = TRUE;
-	}
+	}*/
 
 	static const char* label[] =
 	{
@@ -1661,7 +1814,7 @@ void LLGLState::checkClientArrays(const std::string& msg, U32 data_mask)
 	};
 
 
-	for (S32 j = 0; j < 4; j++)
+	for (S32 j = 1; j < 4; j++)
 	{
 		if (glIsEnabled(value[j]))
 		{
@@ -1783,17 +1936,26 @@ LLGLState::LLGLState(LLGLenum state, S32 enabled) :
 	mState(state), mWasEnabled(FALSE), mIsEnabled(FALSE)
 {
 	if (LLGLSLShader::sNoFixedFunction)
-	{ //always disable state that's deprecated post GL 3.0
+	{ //always ignore state that's deprecated post GL 3.0
 		switch (state)
 		{
 			case GL_ALPHA_TEST:
-				enabled = 0;
+			case GL_NORMALIZE:
+			case GL_TEXTURE_GEN_R:
+			case GL_TEXTURE_GEN_S:
+			case GL_TEXTURE_GEN_T:
+			case GL_TEXTURE_GEN_Q:
+			case GL_LIGHTING:
+			case GL_COLOR_MATERIAL:
+			case GL_FOG:
+			case GL_LINE_STIPPLE:
+				mState = 0;
 				break;
 		}
 	}
 
 	stop_glerror();
-	if (state)
+	if (mState)
 	{
 		mWasEnabled = sStateMap[state];
 		llassert(mWasEnabled == glIsEnabled(state));
@@ -1875,79 +2037,6 @@ void LLGLManager::initGLStates()
 
 ////////////////////////////////////////////////////////////////////////////////
 
-void enable_vertex_weighting(const S32 index)
-{
-#if GL_ARB_vertex_program
-	if (index > 0) glEnableVertexAttribArrayARB(index);	// vertex weights
-#endif
-}
-
-void disable_vertex_weighting(const S32 index)
-{
-#if GL_ARB_vertex_program
-	if (index > 0) glDisableVertexAttribArrayARB(index);	// vertex weights
-#endif
-}
-
-void enable_binormals(const S32 index)
-{
-#if GL_ARB_vertex_program
-	if (index > 0)
-	{
-		glEnableVertexAttribArrayARB(index);	// binormals
-	}
-#endif
-}
-
-void disable_binormals(const S32 index)
-{
-#if GL_ARB_vertex_program
-	if (index > 0)
-	{
-		glDisableVertexAttribArrayARB(index);	// binormals
-	}
-#endif
-}
-
-
-void enable_cloth_weights(const S32 index)
-{
-#if GL_ARB_vertex_program
-	if (index > 0)	glEnableVertexAttribArrayARB(index);
-#endif
-}
-
-void disable_cloth_weights(const S32 index)
-{
-#if GL_ARB_vertex_program
-	if (index > 0) glDisableVertexAttribArrayARB(index);
-#endif
-}
-
-void set_vertex_weights(const S32 index, const U32 stride, const F32 *weights)
-{
-#if GL_ARB_vertex_program
-	if (index > 0) glVertexAttribPointerARB(index, 1, GL_FLOAT, FALSE, stride, weights);
-	stop_glerror();
-#endif
-}
-
-void set_vertex_clothing_weights(const S32 index, const U32 stride, const LLVector4 *weights)
-{
-#if GL_ARB_vertex_program
-	if (index > 0) glVertexAttribPointerARB(index, 4, GL_FLOAT, TRUE, stride, weights);
-	stop_glerror();
-#endif
-}
-
-void set_binormals(const S32 index, const U32 stride,const LLVector3 *binormals)
-{
-#if GL_ARB_vertex_program
-	if (index > 0) glVertexAttribPointerARB(index, 3, GL_FLOAT, FALSE, stride, binormals);
-	stop_glerror();
-#endif
-}
-
 void parse_gl_version( S32* major, S32* minor, S32* release, std::string* vendor_specific )
 {
 	// GL_VERSION returns a null-terminated string with the format: 
@@ -2028,6 +2117,55 @@ void parse_gl_version( S32* major, S32* minor, S32* release, std::string* vendor
 	}
 }
 
+
+void parse_glsl_version(S32& major, S32& minor)
+{
+	// GL_SHADING_LANGUAGE_VERSION returns a null-terminated string with the format: 
+	// <major>.<minor>[.<release>] [<vendor specific>]
+
+	const char* version = (const char*) glGetString(GL_SHADING_LANGUAGE_VERSION);
+	major = 0;
+	minor = 0;
+	
+	if( !version )
+	{
+		return;
+	}
+
+	std::string ver_copy( version );
+	S32 len = (S32)strlen( version );	/* Flawfinder: ignore */
+	S32 i = 0;
+	S32 start;
+	// Find the major version
+	start = i;
+	for( ; i < len; i++ )
+	{
+		if( '.' == version[i] )
+		{
+			break;
+		}
+	}
+	std::string major_str = ver_copy.substr(start,i-start);
+	LLStringUtil::convertToS32(major_str, major);
+
+	if( '.' == version[i] )
+	{
+		i++;
+	}
+
+	// Find the minor version
+	start = i;
+	for( ; i < len; i++ )
+	{
+		if( ('.' == version[i]) || isspace(version[i]) )
+		{
+			break;
+		}
+	}
+	std::string minor_str = ver_copy.substr(start,i-start);
+	LLStringUtil::convertToS32(minor_str, minor);
+}
+
 LLGLUserClipPlane::LLGLUserClipPlane(const LLPlane& p, const glh::matrix4f& modelview, const glh::matrix4f& projection, bool apply)
 {
 	mApply = apply;
@@ -2060,20 +2198,20 @@ void LLGLUserClipPlane::setPlane(F32 a, F32 b, F32 c, F32 d)
     glh::matrix4f suffix;
     suffix.set_row(2, cplane);
     glh::matrix4f newP = suffix * P;
-    glMatrixMode(GL_PROJECTION);
-	glPushMatrix();
-    glLoadMatrixf(newP.m);
+    gGL.matrixMode(LLRender::MM_PROJECTION);
+	gGL.pushMatrix();
+    gGL.loadMatrix(newP.m);
 	gGLObliqueProjectionInverse = LLMatrix4(newP.inverse().transpose().m);
-    glMatrixMode(GL_MODELVIEW);
+    gGL.matrixMode(LLRender::MM_MODELVIEW);
 }
 
 LLGLUserClipPlane::~LLGLUserClipPlane()
 {
 	if (mApply)
 	{
-		glMatrixMode(GL_PROJECTION);
-		glPopMatrix();
-		glMatrixMode(GL_MODELVIEW);
+		gGL.matrixMode(LLRender::MM_PROJECTION);
+		gGL.popMatrix();
+		gGL.matrixMode(LLRender::MM_MODELVIEW);
 	}
 }
 
@@ -2263,16 +2401,16 @@ LLGLSquashToFarClip::LLGLSquashToFarClip(glh::matrix4f P, U32 layer)
 		P.element(2, i) = P.element(3, i) * depth;
 	}
 
-	glMatrixMode(GL_PROJECTION);
-	glPushMatrix();
-	glLoadMatrixf(P.m);
-	glMatrixMode(GL_MODELVIEW);
+	gGL.matrixMode(LLRender::MM_PROJECTION);
+	gGL.pushMatrix();
+	gGL.loadMatrix(P.m);
+	gGL.matrixMode(LLRender::MM_MODELVIEW);
 }
 
 LLGLSquashToFarClip::~LLGLSquashToFarClip()
 {
-	glMatrixMode(GL_PROJECTION);
-	glPopMatrix();
-	glMatrixMode(GL_MODELVIEW);
+	gGL.matrixMode(LLRender::MM_PROJECTION);
+	gGL.popMatrix();
+	gGL.matrixMode(LLRender::MM_MODELVIEW);
 }
 
diff --git a/indra/llrender/llgl.h b/indra/llrender/llgl.h
index d736133f3f..5a33c98708 100644
--- a/indra/llrender/llgl.h
+++ b/indra/llrender/llgl.h
@@ -88,6 +88,7 @@ public:
 		
 	// ARB Extensions
 	BOOL mHasVertexBufferObject;
+	BOOL mHasVertexArrayObject;
 	BOOL mHasSync;
 	BOOL mHasMapBufferRange;
 	BOOL mHasFlushBufferRange;
@@ -112,6 +113,7 @@ public:
 	BOOL mHasAnisotropic;
 	BOOL mHasARBEnvCombine;
 	BOOL mHasCubeMap;
+	BOOL mHasDebugOutput;
 
 	// Vendor-specific extensions
 	BOOL mIsATI;
@@ -136,6 +138,8 @@ public:
 	S32 mDriverVersionMinor;
 	S32 mDriverVersionRelease;
 	F32 mGLVersion; // e.g = 1.4
+	S32 mGLSLVersionMajor;
+	S32 mGLSLVersionMinor;
 	std::string mDriverVersionVendorString;
 
 	S32 mVRAM; // VRAM in MB
@@ -148,7 +152,6 @@ public:
 	void printGLInfoString();
 	void getGLInfo(LLSD& info);
 
-	U32 getNumFBOFSAASamples(U32 desired_samples = 32);
 	// In ALL CAPS
 	std::string mGLVendor;
 	std::string mGLVendorShort;
@@ -252,7 +255,7 @@ public:
 	static void dumpStates();
 	static void checkStates(const std::string& msg = "");
 	static void checkTextureChannels(const std::string& msg = "");
-	static void checkClientArrays(const std::string& msg = "", U32 data_mask = 0x0001);
+	static void checkClientArrays(const std::string& msg = "", U32 data_mask = 0);
 	
 protected:
 	static boost::unordered_map<LLGLenum, LLGLboolean> sStateMap;
@@ -419,15 +422,7 @@ extern LLMatrix4 gGLObliqueProjectionInverse;
 #include "llglstates.h"
 
 void init_glstates();
-void enable_vertex_weighting(const S32 index);
-void disable_vertex_weighting(const S32 index);
-void enable_binormals(const S32 index);
-void disable_binormals(const S32 index);
-void enable_cloth_weights(const S32 index);
-void disable_cloth_weights(const S32 index);
-void set_vertex_weights(const S32 index, const U32 stride, const F32 *weights);
-void set_vertex_clothing_weights(const S32 index, const U32 stride, const LLVector4 *weights);
-void set_binormals(const S32 index, const U32 stride, const LLVector3 *binormals);
+
 void parse_gl_version( S32* major, S32* minor, S32* release, std::string* vendor_specific );
 
 extern BOOL gClothRipple;
diff --git a/indra/llrender/llglheaders.h b/indra/llrender/llglheaders.h
index 851a75629e..d61ec707f0 100644
--- a/indra/llrender/llglheaders.h
+++ b/indra/llrender/llglheaders.h
@@ -68,6 +68,12 @@ extern PFNGLUNMAPBUFFERARBPROC		glUnmapBufferARB;
 extern PFNGLGETBUFFERPARAMETERIVARBPROC	glGetBufferParameterivARB;
 extern PFNGLGETBUFFERPOINTERVARBPROC	glGetBufferPointervARB;
 
+// GL_ARB_vertex_array_object
+extern PFNGLBINDVERTEXARRAYPROC glBindVertexArray;
+extern PFNGLDELETEVERTEXARRAYSPROC glDeleteVertexArrays;
+extern PFNGLGENVERTEXARRAYSPROC glGenVertexArrays;
+extern PFNGLISVERTEXARRAYPROC glIsVertexArray;
+
 // GL_ARB_sync
 extern PFNGLFENCESYNCPROC				glFenceSync;
 extern PFNGLISSYNCPROC					glIsSync;
@@ -193,6 +199,7 @@ extern PFNGLVERTEXATTRIB4UBVARBPROC glVertexAttrib4ubvARB;
 extern PFNGLVERTEXATTRIB4UIVARBPROC glVertexAttrib4uivARB;
 extern PFNGLVERTEXATTRIB4USVARBPROC glVertexAttrib4usvARB;
 extern PFNGLVERTEXATTRIBPOINTERARBPROC glVertexAttribPointerARB;
+extern PFNGLVERTEXATTRIBIPOINTERPROC glVertexAttribIPointer;
 extern PFNGLENABLEVERTEXATTRIBARRAYARBPROC glEnableVertexAttribArrayARB;
 extern PFNGLDISABLEVERTEXATTRIBARRAYARBPROC glDisableVertexAttribArrayARB;
 extern PFNGLPROGRAMSTRINGARBPROC glProgramStringARB;
@@ -310,6 +317,12 @@ extern PFNGLCLIENTACTIVETEXTUREARBPROC	glClientActiveTextureARB;
 extern PFNGLDRAWRANGEELEMENTSPROC 	glDrawRangeElements;
 #endif // LL_LINUX_NV_GL_HEADERS
 
+// GL_ARB_vertex_array_object
+extern PFNGLBINDVERTEXARRAYPROC glBindVertexArray;
+extern PFNGLDELETEVERTEXARRAYSPROC glDeleteVertexArrays;
+extern PFNGLGENVERTEXARRAYSPROC glGenVertexArrays;
+extern PFNGLISVERTEXARRAYPROC glIsVertexArray;
+
 // GL_ARB_vertex_buffer_object
 extern PFNGLBINDBUFFERARBPROC		glBindBufferARB;
 extern PFNGLDELETEBUFFERSARBPROC	glDeleteBuffersARB;
@@ -448,6 +461,7 @@ extern PFNGLVERTEXATTRIB4UBVARBPROC glVertexAttrib4ubvARB;
 extern PFNGLVERTEXATTRIB4UIVARBPROC glVertexAttrib4uivARB;
 extern PFNGLVERTEXATTRIB4USVARBPROC glVertexAttrib4usvARB;
 extern PFNGLVERTEXATTRIBPOINTERARBPROC glVertexAttribPointerARB;
+extern PFNGLVERTEXATTRIBIPOINTERPROC glVertexAttribIPointer;
 extern PFNGLENABLEVERTEXATTRIBARRAYARBPROC glEnableVertexAttribArrayARB;
 extern PFNGLDISABLEVERTEXATTRIBARRAYARBPROC glDisableVertexAttribArrayARB;
 extern PFNGLPROGRAMSTRINGARBPROC glProgramStringARB;
@@ -531,6 +545,9 @@ extern PFNGLSAMPLEMASKIPROC glSampleMaski;
 #include "GL/glext.h"
 #include "GL/glh_extensions.h"
 
+// WGL_ARB_create_context
+extern PFNWGLCREATECONTEXTATTRIBSARBPROC wglCreateContextAttribsARB;
+extern PFNGLGETSTRINGIPROC glGetStringi;
 
 // GL_ARB_vertex_buffer_object
 extern PFNGLBINDBUFFERARBPROC		glBindBufferARB;
@@ -545,6 +562,12 @@ extern PFNGLUNMAPBUFFERARBPROC		glUnmapBufferARB;
 extern PFNGLGETBUFFERPARAMETERIVARBPROC	glGetBufferParameterivARB;
 extern PFNGLGETBUFFERPOINTERVARBPROC	glGetBufferPointervARB;
 
+// GL_ARB_vertex_array_object
+extern PFNGLBINDVERTEXARRAYPROC glBindVertexArray;
+extern PFNGLDELETEVERTEXARRAYSPROC glDeleteVertexArrays;
+extern PFNGLGENVERTEXARRAYSPROC glGenVertexArrays;
+extern PFNGLISVERTEXARRAYPROC glIsVertexArray;
+
 // GL_ARB_sync
 extern PFNGLFENCESYNCPROC				glFenceSync;
 extern PFNGLISSYNCPROC					glIsSync;
@@ -672,6 +695,7 @@ extern PFNGLVERTEXATTRIB4UBVARBPROC glVertexAttrib4ubvARB;
 extern PFNGLVERTEXATTRIB4UIVARBPROC glVertexAttrib4uivARB;
 extern PFNGLVERTEXATTRIB4USVARBPROC glVertexAttrib4usvARB;
 extern PFNGLVERTEXATTRIBPOINTERARBPROC glVertexAttribPointerARB;
+extern PFNGLVERTEXATTRIBIPOINTERPROC glVertexAttribIPointer;
 extern PFNGLENABLEVERTEXATTRIBARRAYARBPROC glEnableVertexAttribArrayARB;
 extern PFNGLDISABLEVERTEXATTRIBARRAYARBPROC glDisableVertexAttribArrayARB;
 extern PFNGLPROGRAMSTRINGARBPROC glProgramStringARB;
@@ -735,6 +759,12 @@ extern PFNGLTEXIMAGE3DMULTISAMPLEPROC glTexImage3DMultisample;
 extern PFNGLGETMULTISAMPLEFVPROC glGetMultisamplefv;
 extern PFNGLSAMPLEMASKIPROC glSampleMaski;
 
+//GL_ARB_debug_output
+extern PFNGLDEBUGMESSAGECONTROLARBPROC glDebugMessageControlARB;
+extern PFNGLDEBUGMESSAGEINSERTARBPROC glDebugMessageInsertARB;
+extern PFNGLDEBUGMESSAGECALLBACKARBPROC glDebugMessageCallbackARB;
+extern PFNGLGETDEBUGMESSAGELOGARBPROC glGetDebugMessageLogARB;
+
 #elif LL_DARWIN
 //----------------------------------------------------------------------------
 // LL_DARWIN
@@ -899,6 +929,31 @@ extern void glGetBufferPointervARB (GLenum, GLenum, GLvoid* *);
 #endif /* GL_GLEXT_FUNCTION_POINTERS */
 #endif
 
+#ifndef GL_ARB_texture_rg
+#define GL_RG                             0x8227
+#define GL_RG_INTEGER                     0x8228
+#define GL_R8                             0x8229
+#define GL_R16                            0x822A
+#define GL_RG8                            0x822B
+#define GL_RG16                           0x822C
+#define GL_R16F                           0x822D
+#define GL_R32F                           0x822E
+#define GL_RG16F                          0x822F
+#define GL_RG32F                          0x8230
+#define GL_R8I                            0x8231
+#define GL_R8UI                           0x8232
+#define GL_R16I                           0x8233
+#define GL_R16UI                          0x8234
+#define GL_R32I                           0x8235
+#define GL_R32UI                          0x8236
+#define GL_RG8I                           0x8237
+#define GL_RG8UI                          0x8238
+#define GL_RG16I                          0x8239
+#define GL_RG16UI                         0x823A
+#define GL_RG32I                          0x823B
+#define GL_RG32UI                         0x823C
+#endif
+
 // May be needed for DARWIN...
 // #ifndef GL_ARB_compressed_tex_image
 // #define GL_ARB_compressed_tex_image 1
diff --git a/indra/llrender/llglslshader.cpp b/indra/llrender/llglslshader.cpp
index f51d83abe4..4b7e639aed 100644
--- a/indra/llrender/llglslshader.cpp
+++ b/indra/llrender/llglslshader.cpp
@@ -31,6 +31,7 @@
 #include "llshadermgr.h"
 #include "llfile.h"
 #include "llrender.h"
+#include "llvertexbuffer.h"
 
 #if LL_DARWIN
 #include "OpenGL/OpenGL.h"
@@ -50,6 +51,7 @@ using std::string;
 
 GLhandleARB LLGLSLShader::sCurBoundShader = 0;
 LLGLSLShader* LLGLSLShader::sCurBoundShaderPtr = NULL;
+S32 LLGLSLShader::sIndexedTextureChannels = 0;
 bool LLGLSLShader::sNoFixedFunction = false;
 
 //UI shader -- declared here so llui_libtest will link properly
@@ -62,10 +64,23 @@ BOOL shouldChange(const LLVector4& v1, const LLVector4& v2)
 }
 
 LLShaderFeatures::LLShaderFeatures()
-: calculatesLighting(false), isShiny(false), isFullbright(false), hasWaterFog(false),
-hasTransport(false), hasSkinning(false), hasObjectSkinning(false), hasAtmospherics(false), isSpecular(false),
-hasGamma(false), hasLighting(false), calculatesAtmospherics(false), mIndexedTextureChannels(0), disableTextureIndex(false),
-hasAlphaMask(false)
+	: atmosphericHelpers(false)
+	, calculatesLighting(false)
+	, calculatesAtmospherics(false)
+	, hasLighting(false)
+	, isAlphaLighting(false)
+	, isShiny(false)
+	, isFullbright(false)
+	, isSpecular(false)
+	, hasWaterFog(false)
+	, hasTransport(false)
+	, hasSkinning(false)
+	, hasObjectSkinning(false)
+	, hasAtmospherics(false)
+	, hasGamma(false)
+	, mIndexedTextureChannels(0)
+	, disableTextureIndex(false)
+	, hasAlphaMask(false)
 {
 }
 
@@ -75,6 +90,7 @@ hasAlphaMask(false)
 LLGLSLShader::LLGLSLShader()
 	: mProgramObject(0), mActiveTextureChannels(0), mShaderLevel(0), mShaderGroup(SG_DEFAULT), mUniformsDirty(FALSE)
 {
+
 }
 
 void LLGLSLShader::unload()
@@ -93,7 +109,12 @@ void LLGLSLShader::unload()
 		glGetAttachedObjectsARB(mProgramObject, 1024, &count, obj);
 		for (GLsizei i = 0; i < count; i++)
 		{
-			glDeleteObjectARB(obj[i]);
+#if !LL_DARWIN
+			if (glIsProgramARB(obj[i]))
+#endif
+			{
+				glDeleteObjectARB(obj[i]);
+			}
 		}
 
 		glDeleteObjectARB(mProgramObject);
@@ -110,17 +131,19 @@ void LLGLSLShader::unload()
 BOOL LLGLSLShader::createShader(vector<string> * attributes,
 								vector<string> * uniforms)
 {
+	//reloading, reset matrix hash values
+	for (U32 i = 0; i < LLRender::NUM_MATRIX_MODES; ++i)
+	{
+		mMatHash[i] = 0xFFFFFFFF;
+	}
+	mLightHash = 0xFFFFFFFF;
+
 	llassert_always(!mShaderFiles.empty());
 	BOOL success = TRUE;
 
 	// Create program
 	mProgramObject = glCreateProgramObjectARB();
 	
-	if (gGLManager.mGLVersion < 3.1f)
-	{ //force indexed texture channels to 1 if GL version is old (performance improvement for drivers with poor branching shader model support)
-		mFeatures.mIndexedTextureChannels = llmin(mFeatures.mIndexedTextureChannels, 1);
-	}
-
 	//compile new source
 	vector< pair<string,GLenum> >::iterator fileIter = mShaderFiles.begin();
 	for ( ; fileIter != mShaderFiles.end(); fileIter++ )
@@ -143,8 +166,9 @@ BOOL LLGLSLShader::createShader(vector<string> * attributes,
 		return FALSE;
 	}
 
-	if (gGLManager.mGLVersion < 3.1f)
-	{ //attachShaderFeatures may have set the number of indexed texture channels, so set to 1 again
+	if (gGLManager.mGLSLVersionMajor < 2 && gGLManager.mGLSLVersionMinor < 3)
+	{ //indexed texture rendering requires GLSL 1.3 or later
+		//attachShaderFeatures may have set the number of indexed texture channels, so set to 1 again
 		mFeatures.mIndexedTextureChannels = llmin(mFeatures.mIndexedTextureChannels, 1);
 	}
 
@@ -235,6 +259,13 @@ void LLGLSLShader::attachObjects(GLhandleARB* objects, S32 count)
 
 BOOL LLGLSLShader::mapAttributes(const vector<string> * attributes)
 {
+	//before linking, make sure reserved attributes always have consistent locations
+	for (U32 i = 0; i < LLShaderMgr::instance()->mReservedAttribs.size(); i++)
+	{
+		const char* name = LLShaderMgr::instance()->mReservedAttribs[i].c_str();
+		glBindAttribLocationARB(mProgramObject, i, (const GLcharARB *) name);
+	}
+	
 	//link the program
 	BOOL res = link();
 
@@ -308,7 +339,7 @@ void LLGLSLShader::mapUniform(GLint index, const vector<string> * uniforms)
 		for (S32 i = 0; i < (S32) LLShaderMgr::instance()->mReservedUniforms.size(); i++)
 		{
 			if ( (mUniform[i] == -1)
-				&& (LLShaderMgr::instance()->mReservedUniforms[i].compare(0, length, name, LLShaderMgr::instance()->mReservedUniforms[i].length()) == 0))
+				&& (LLShaderMgr::instance()->mReservedUniforms[i] == name))
 			{
 				//found it
 				mUniform[i] = location;
@@ -322,7 +353,7 @@ void LLGLSLShader::mapUniform(GLint index, const vector<string> * uniforms)
 			for (U32 i = 0; i < uniforms->size(); i++)
 			{
 				if ( (mUniform[i+LLShaderMgr::instance()->mReservedUniforms.size()] == -1)
-					&& ((*uniforms)[i].compare(0, length, name, (*uniforms)[i].length()) == 0))
+					&& ((*uniforms)[i] == name))
 				{
 					//found it
 					mUniform[i+LLShaderMgr::instance()->mReservedUniforms.size()] = location;
@@ -386,6 +417,7 @@ void LLGLSLShader::bind()
 	gGL.flush();
 	if (gGLManager.mHasShaderObjects)
 	{
+		LLVertexBuffer::unbind();
 		glUseProgramObjectARB(mProgramObject);
 		sCurBoundShader = mProgramObject;
 		sCurBoundShaderPtr = this;
@@ -411,6 +443,7 @@ void LLGLSLShader::unbind()
 				stop_glerror();
 			}
 		}
+		LLVertexBuffer::unbind();
 		glUseProgramObjectARB(0);
 		sCurBoundShader = 0;
 		sCurBoundShaderPtr = NULL;
@@ -420,9 +453,13 @@ void LLGLSLShader::unbind()
 
 void LLGLSLShader::bindNoShader(void)
 {
-	glUseProgramObjectARB(0);
-	sCurBoundShader = 0;
-	sCurBoundShaderPtr = NULL;
+	LLVertexBuffer::unbind();
+	if (gGLManager.mHasShaderObjects)
+	{
+		glUseProgramObjectARB(0);
+		sCurBoundShader = 0;
+		sCurBoundShaderPtr = NULL;
+	}
 }
 
 S32 LLGLSLShader::enableTexture(S32 uniform, LLTexUnit::eTextureType mode)
@@ -768,13 +805,17 @@ GLint LLGLSLShader::getUniformLocation(const string& uniform)
 		}
 	}
 
-	/*if (gDebugGL)
+	return ret;
+}
+
+GLint LLGLSLShader::getUniformLocation(U32 index)
+{
+	GLint ret = -1;
+	if (mProgramObject > 0)
 	{
-		if (ret == -1 && ret != glGetUniformLocationARB(mProgramObject, uniform.c_str()))
-		{
-			llerrs << "Uniform map invalid." << llendl;
-		}
-	}*/
+		llassert(index < mUniform.size());
+		return mUniform[index];
+	}
 
 	return ret;
 }
@@ -930,7 +971,9 @@ void LLGLSLShader::uniform4fv(const string& uniform, U32 count, const GLfloat* v
 		std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
 		if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
 		{
+			stop_glerror();
 			glUniform4fvARB(location, count, v);
+			stop_glerror();
 			mValue[location] = vec;
 		}
 	}
@@ -985,8 +1028,8 @@ void LLGLSLShader::vertexAttrib4fv(U32 index, GLfloat* v)
 	}
 }
 
-void LLGLSLShader::setAlphaRange(F32 minimum, F32 maximum)
+void LLGLSLShader::setMinimumAlpha(F32 minimum)
 {
-	uniform1f("minimum_alpha", minimum);
-	uniform1f("maximum_alpha", maximum);
+	gGL.flush();
+	uniform1f(LLShaderMgr::MINIMUM_ALPHA, minimum);
 }
diff --git a/indra/llrender/llglslshader.h b/indra/llrender/llglslshader.h
index 558ea66b50..7873fe3c4e 100644
--- a/indra/llrender/llglslshader.h
+++ b/indra/llrender/llglslshader.h
@@ -33,9 +33,11 @@
 class LLShaderFeatures
 {
 public:
+	bool atmosphericHelpers;
 	bool calculatesLighting;
 	bool calculatesAtmospherics;
 	bool hasLighting; // implies no transport (it's possible to have neither though)
+	bool isAlphaLighting; // indicates lighting shaders need not be linked in (lighting performed directly in alpha shader to match deferred lighting functions)
 	bool isShiny;
 	bool isFullbright; // implies no lighting
 	bool isSpecular;
@@ -69,7 +71,7 @@ public:
 
 	static GLhandleARB sCurBoundShader;
 	static LLGLSLShader* sCurBoundShaderPtr;
-
+	static S32 sIndexedTextureChannels;
 	static bool sNoFixedFunction;
 
 	void unload();
@@ -108,16 +110,17 @@ public:
 	void uniformMatrix3fv(const std::string& uniform, U32 count, GLboolean transpose, const GLfloat *v);
 	void uniformMatrix4fv(const std::string& uniform, U32 count, GLboolean transpose, const GLfloat *v);
 
-	void setAlphaRange(F32 minimum, F32 maximum);
+	void setMinimumAlpha(F32 minimum);
 
 	void vertexAttrib4f(U32 index, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
 	void vertexAttrib4fv(U32 index, GLfloat* v);
 	
 	GLint getUniformLocation(const std::string& uniform);
+	GLint getUniformLocation(U32 index);
+
 	GLint getAttribLocation(U32 attrib);
 	GLint mapUniformTextureChannel(GLint location, GLenum type);
 	
-
 	//enable/disable texture channel for specified uniform
 	//if given texture uniform is active in the shader, 
 	//the corresponding channel will be active upon return
@@ -132,6 +135,9 @@ public:
 	// Unbinds any previously bound shader by explicitly binding no shader.
 	static void bindNoShader(void);
 
+	U32 mMatHash[LLRender::NUM_MATRIX_MODES];
+	U32 mLightHash;
+
 	GLhandleARB mProgramObject;
 	std::vector<GLint> mAttribute; //lookup table of attribute enum to attribute channel
 	std::vector<GLint> mUniform;   //lookup table of uniform enum to uniform location
diff --git a/indra/llrender/llimagegl.cpp b/indra/llrender/llimagegl.cpp
index 7188b0fa44..78591ddd38 100644
--- a/indra/llrender/llimagegl.cpp
+++ b/indra/llrender/llimagegl.cpp
@@ -36,7 +36,9 @@
 
 #include "llmath.h"
 #include "llgl.h"
+#include "llglslshader.h"
 #include "llrender.h"
+
 //----------------------------------------------------------------------------
 const F32 MIN_TEXTURE_LIFETIME = 10.f;
 
@@ -725,7 +727,10 @@ void LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 		{
 			if (mAutoGenMips)
 			{
-				glTexParameteri(LLTexUnit::getInternalType(mBindTarget), GL_GENERATE_MIPMAP_SGIS, TRUE);
+				if (!gGLManager.mHasFramebufferObject)
+				{
+					glTexParameteri(LLTexUnit::getInternalType(mBindTarget), GL_GENERATE_MIPMAP_SGIS, TRUE);
+				}
 				stop_glerror();
 				{
 // 					LLFastTimer t2(FTM_TEMP4);
@@ -754,6 +759,11 @@ void LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 						stop_glerror();
 					}
 				}
+
+				if (gGLManager.mHasFramebufferObject)
+				{
+					glGenerateMipmap(LLTexUnit::getInternalType(mBindTarget));
+				}
 			}
 			else
 			{
@@ -875,6 +885,9 @@ void LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 
 BOOL LLImageGL::preAddToAtlas(S32 discard_level, const LLImageRaw* raw_image)
 {
+	//not compatible with core GL profile
+	llassert(!LLRender::sGLCoreProfile);
+
 	if (gGLManager.mIsDisabled)
 	{
 		llwarns << "Trying to create a texture while GL is disabled!" << llendl;
@@ -901,29 +914,29 @@ BOOL LLImageGL::preAddToAtlas(S32 discard_level, const LLImageRaw* raw_image)
 	{
 		switch (mComponents)
 		{
-		  case 1:
+			case 1:
 			// Use luminance alpha (for fonts)
 			mFormatInternal = GL_LUMINANCE8;
 			mFormatPrimary = GL_LUMINANCE;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  case 2:
+			case 2:
 			// Use luminance alpha (for fonts)
 			mFormatInternal = GL_LUMINANCE8_ALPHA8;
 			mFormatPrimary = GL_LUMINANCE_ALPHA;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  case 3:
+			case 3:
 			mFormatInternal = GL_RGB8;
 			mFormatPrimary = GL_RGB;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  case 4:
+			case 4:
 			mFormatInternal = GL_RGBA8;
 			mFormatPrimary = GL_RGBA;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  default:
+			default:
 			llerrs << "Bad number of components for texture: " << (U32)getComponents() << llendl;
 		}
 	}
@@ -1099,8 +1112,75 @@ void LLImageGL::deleteTextures(S32 numTextures, U32 *textures, bool immediate)
 // static
 void LLImageGL::setManualImage(U32 target, S32 miplevel, S32 intformat, S32 width, S32 height, U32 pixformat, U32 pixtype, const void *pixels)
 {
-	glTexImage2D(target, miplevel, intformat, width, height, 0, pixformat, pixtype, pixels);
+	bool use_scratch = false;
+	U32* scratch = NULL;
+	if (LLRender::sGLCoreProfile)
+	{
+		if (pixformat == GL_ALPHA && pixtype == GL_UNSIGNED_BYTE) 
+		{ //GL_ALPHA is deprecated, convert to RGBA
+			use_scratch = true;
+			scratch = new U32[width*height];
+
+			U32 pixel_count = (U32) (width*height);
+			for (U32 i = 0; i < pixel_count; i++)
+			{
+				U8* pix = (U8*) &scratch[i];
+				pix[0] = pix[1] = pix[2] = 0;
+				pix[3] = ((U8*) pixels)[i];
+			}				
+			
+			pixformat = GL_RGBA;
+			intformat = GL_RGBA8;
+		}
+
+		if (pixformat == GL_LUMINANCE_ALPHA && pixtype == GL_UNSIGNED_BYTE) 
+		{ //GL_LUMINANCE_ALPHA is deprecated, convert to RGBA
+			use_scratch = true;
+			scratch = new U32[width*height];
+
+			U32 pixel_count = (U32) (width*height);
+			for (U32 i = 0; i < pixel_count; i++)
+			{
+				U8 lum = ((U8*) pixels)[i*2+0];
+				U8 alpha = ((U8*) pixels)[i*2+1];
+
+				U8* pix = (U8*) &scratch[i];
+				pix[0] = pix[1] = pix[2] = lum;
+				pix[3] = alpha;
+			}				
+			
+			pixformat = GL_RGBA;
+			intformat = GL_RGBA8;
+		}
+
+		if (pixformat == GL_LUMINANCE && pixtype == GL_UNSIGNED_BYTE) 
+		{ //GL_LUMINANCE_ALPHA is deprecated, convert to RGB
+			use_scratch = true;
+			scratch = new U32[width*height];
+
+			U32 pixel_count = (U32) (width*height);
+			for (U32 i = 0; i < pixel_count; i++)
+			{
+				U8 lum = ((U8*) pixels)[i];
+				
+				U8* pix = (U8*) &scratch[i];
+				pix[0] = pix[1] = pix[2] = lum;
+				pix[3] = 255;
+			}				
+			
+			pixformat = GL_RGBA;
+			intformat = GL_RGB8;
+		}
+	}
+
 	stop_glerror();
+	glTexImage2D(target, miplevel, intformat, width, height, 0, pixformat, pixtype, use_scratch ? scratch : pixels);
+	stop_glerror();
+
+	if (use_scratch)
+	{
+		delete [] scratch;
+	}
 }
 
 //create an empty GL texture: just create a texture name
@@ -1167,29 +1247,29 @@ BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S
 	{
 		switch (mComponents)
 		{
-		  case 1:
+			case 1:
 			// Use luminance alpha (for fonts)
 			mFormatInternal = GL_LUMINANCE8;
 			mFormatPrimary = GL_LUMINANCE;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  case 2:
+			case 2:
 			// Use luminance alpha (for fonts)
 			mFormatInternal = GL_LUMINANCE8_ALPHA8;
 			mFormatPrimary = GL_LUMINANCE_ALPHA;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  case 3:
+			case 3:
 			mFormatInternal = GL_RGB8;
 			mFormatPrimary = GL_RGB;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  case 4:
+			case 4:
 			mFormatInternal = GL_RGBA8;
 			mFormatPrimary = GL_RGBA;
 			mFormatType = GL_UNSIGNED_BYTE;
 			break;
-		  default:
+			default:
 			llerrs << "Bad number of components for texture: " << (U32)getComponents() << llendl;
 		}
 
@@ -1212,6 +1292,7 @@ BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S
 BOOL LLImageGL::createGLTexture(S32 discard_level, const U8* data_in, BOOL data_hasmips, S32 usename)
 {
 	llassert(data_in);
+	stop_glerror();
 
 	if (discard_level < 0)
 	{
@@ -1240,8 +1321,11 @@ BOOL LLImageGL::createGLTexture(S32 discard_level, const U8* data_in, BOOL data_
 		stop_glerror();
 		{
 			llverify(gGL.getTexUnit(0)->bind(this));
+			stop_glerror();
 			glTexParameteri(LLTexUnit::getInternalType(mBindTarget), GL_TEXTURE_BASE_LEVEL, 0);
+			stop_glerror();
 			glTexParameteri(LLTexUnit::getInternalType(mBindTarget), GL_TEXTURE_MAX_LEVEL,  mMaxDiscardLevel-discard_level);
+			stop_glerror();
 		}
 	}
 	if (!mTexName)
@@ -1754,7 +1838,7 @@ void LLImageGL::analyzeAlpha(const void* data_in, U32 w, U32 h)
 	// this to be an intentional effect and don't treat as a mask.
 
 	U32 midrangetotal = 0;
-	for (U32 i = 4; i < 11; i++)
+	for (U32 i = 2; i < 13; i++)
 	{
 		midrangetotal += sample[i];
 	}
@@ -1769,7 +1853,7 @@ void LLImageGL::analyzeAlpha(const void* data_in, U32 w, U32 h)
 		upperhalftotal += sample[i];
 	}
 
-	if (midrangetotal > length/16 || // lots of midrange, or
+	if (midrangetotal > length/48 || // lots of midrange, or
 	    (lowerhalftotal == length && alphatotal != 0) || // all close to transparent but not all totally transparent, or
 	    (upperhalftotal == length && alphatotal != 255*length)) // all close to opaque but not all totally opaque
 	{
diff --git a/indra/llrender/llpostprocess.cpp b/indra/llrender/llpostprocess.cpp
index d76b2d9004..c0045c8044 100644
--- a/indra/llrender/llpostprocess.cpp
+++ b/indra/llrender/llpostprocess.cpp
@@ -466,21 +466,21 @@ void LLPostProcess::drawOrthoQuad(unsigned int width, unsigned int height, QuadT
 
 void LLPostProcess::viewOrthogonal(unsigned int width, unsigned int height)
 {
-	glMatrixMode(GL_PROJECTION);
-	glPushMatrix();
-	glLoadIdentity();
-	glOrtho( 0.f, (GLdouble) width , (GLdouble) height , 0.f, -1.f, 1.f );
-	glMatrixMode(GL_MODELVIEW);
-	glPushMatrix();
-	glLoadIdentity();
+	gGL.matrixMode(LLRender::MM_PROJECTION);
+	gGL.pushMatrix();
+	gGL.loadIdentity();
+	gGL.ortho( 0.f, (GLdouble) width , (GLdouble) height , 0.f, -1.f, 1.f );
+	gGL.matrixMode(LLRender::MM_MODELVIEW);
+	gGL.pushMatrix();
+	gGL.loadIdentity();
 }
 
 void LLPostProcess::viewPerspective(void)
 {
-	glMatrixMode( GL_PROJECTION );
-	glPopMatrix();
-	glMatrixMode( GL_MODELVIEW );
-	glPopMatrix();
+	gGL.matrixMode(LLRender::MM_PROJECTION);
+	gGL.popMatrix();
+	gGL.matrixMode(LLRender::MM_MODELVIEW);
+	gGL.popMatrix();
 }
 
 void LLPostProcess::changeOrthogonal(unsigned int width, unsigned int height)
diff --git a/indra/llrender/llrender.cpp b/indra/llrender/llrender.cpp
index edcc47aa14..b0ddacbb05 100644
--- a/indra/llrender/llrender.cpp
+++ b/indra/llrender/llrender.cpp
@@ -1,4 +1,4 @@
-/** 
+ /** 
  * @file llrender.cpp
  * @brief LLRender implementation
  *
@@ -34,19 +34,21 @@
 #include "llimagegl.h"
 #include "llrendertarget.h"
 #include "lltexture.h"
+#include "llshadermgr.h"
 
 LLRender gGL;
 
 // Handy copies of last good GL matrices
-F64	gGLModelView[16];
-F64	gGLLastModelView[16];
-F64 gGLLastProjection[16];
-F64 gGLProjection[16];
+F32	gGLModelView[16];
+F32	gGLLastModelView[16];
+F32 gGLLastProjection[16];
+F32 gGLProjection[16];
 S32	gGLViewport[4];
 
 U32 LLRender::sUICalls = 0;
 U32 LLRender::sUIVerts = 0;
 U32 LLTexUnit::sWhiteTexture = 0;
+bool LLRender::sGLCoreProfile = false;
 
 static const U32 LL_NUM_TEXTURE_LAYERS = 32; 
 static const U32 LL_NUM_LIGHT_UNITS = 8;
@@ -178,10 +180,13 @@ void LLTexUnit::enable(eTextureType type)
 
 	if ( (mCurrTexType != type || gGL.mDirty) && (type != TT_NONE) )
 	{
+		stop_glerror();
 		activate();
+		stop_glerror();
 		if (mCurrTexType != TT_NONE && !gGL.mDirty)
 		{
 			disable(); // Force a disable of a previous texture type if it's enabled.
+			stop_glerror();
 		}
 		mCurrTexType = type;
 
@@ -190,7 +195,9 @@ void LLTexUnit::enable(eTextureType type)
 			type != LLTexUnit::TT_MULTISAMPLE_TEXTURE &&
 			mIndex < gGLManager.mNumTextureUnits)
 		{
+			stop_glerror();
 			glEnable(sGLTextureType[type]);
+			stop_glerror();
 		}
 	}
 }
@@ -286,26 +293,35 @@ bool LLTexUnit::bind(LLImageGL* texture, bool for_rendering, bool forceBind)
 		{
 			return bind(LLImageGL::sDefaultGLTexture) ;
 		}
+		stop_glerror();
 		return false ;
 	}
 
 	if ((mCurrTexture != texture->getTexName()) || forceBind)
 	{
 		gGL.flush();
+		stop_glerror();
 		activate();
+		stop_glerror();
 		enable(texture->getTarget());
+		stop_glerror();
 		mCurrTexture = texture->getTexName();
 		glBindTexture(sGLTextureType[texture->getTarget()], mCurrTexture);
+		stop_glerror();
 		texture->updateBindStats(texture->mTextureMemory);		
 		mHasMipMaps = texture->mHasMipMaps;
 		if (texture->mTexOptionsDirty)
 		{
+			stop_glerror();
 			texture->mTexOptionsDirty = false;
 			setTextureAddressMode(texture->mAddressMode);
 			setTextureFilteringOption(texture->mFilterOption);
+			stop_glerror();
 		}
 	}
 
+	stop_glerror();
+
 	return true;
 }
 
@@ -814,14 +830,16 @@ LLLightState::LLLightState(S32 index)
 	mAmbient.set(0,0,0,1);
 	mPosition.set(0,0,1,0);
 	mSpotDirection.set(0,0,-1);
-
 }
 
 void LLLightState::enable()
 {
 	if (!mEnabled)
 	{
-		glEnable(GL_LIGHT0+mIndex);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glEnable(GL_LIGHT0+mIndex);
+		}
 		mEnabled = true;
 	}
 }
@@ -830,7 +848,10 @@ void LLLightState::disable()
 {
 	if (mEnabled)
 	{
-		glDisable(GL_LIGHT0+mIndex);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glDisable(GL_LIGHT0+mIndex);
+		}
 		mEnabled = false;
 	}
 }
@@ -839,8 +860,12 @@ void LLLightState::setDiffuse(const LLColor4& diffuse)
 {
 	if (mDiffuse != diffuse)
 	{
+		++gGL.mLightHash;
 		mDiffuse = diffuse;
-		glLightfv(GL_LIGHT0+mIndex, GL_DIFFUSE, mDiffuse.mV);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightfv(GL_LIGHT0+mIndex, GL_DIFFUSE, mDiffuse.mV);
+		}
 	}
 }
 
@@ -848,8 +873,12 @@ void LLLightState::setAmbient(const LLColor4& ambient)
 {
 	if (mAmbient != ambient)
 	{
+		++gGL.mLightHash;
 		mAmbient = ambient;
-		glLightfv(GL_LIGHT0+mIndex, GL_AMBIENT, mAmbient.mV);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightfv(GL_LIGHT0+mIndex, GL_AMBIENT, mAmbient.mV);
+		}
 	}
 }
 
@@ -857,16 +886,34 @@ void LLLightState::setSpecular(const LLColor4& specular)
 {
 	if (mSpecular != specular)
 	{
+		++gGL.mLightHash;
 		mSpecular = specular;
-		glLightfv(GL_LIGHT0+mIndex, GL_SPECULAR, mSpecular.mV);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightfv(GL_LIGHT0+mIndex, GL_SPECULAR, mSpecular.mV);
+		}
 	}
 }
 
 void LLLightState::setPosition(const LLVector4& position)
 {
 	//always set position because modelview matrix may have changed
+	++gGL.mLightHash;
 	mPosition = position;
-	glLightfv(GL_LIGHT0+mIndex, GL_POSITION, mPosition.mV);
+	if (!LLGLSLShader::sNoFixedFunction)
+	{
+		glLightfv(GL_LIGHT0+mIndex, GL_POSITION, mPosition.mV);
+	}
+	else
+	{ //transform position by current modelview matrix
+		glh::vec4f pos(position.mV);
+
+		const glh::matrix4f& mat = gGL.getModelviewMatrix();
+		mat.mult_matrix_vec(pos);
+
+		mPosition.set(pos.v);
+	}
+
 }
 
 void LLLightState::setConstantAttenuation(const F32& atten)
@@ -874,7 +921,11 @@ void LLLightState::setConstantAttenuation(const F32& atten)
 	if (mConstantAtten != atten)
 	{
 		mConstantAtten = atten;
-		glLightf(GL_LIGHT0+mIndex, GL_CONSTANT_ATTENUATION, atten);
+		++gGL.mLightHash;
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightf(GL_LIGHT0+mIndex, GL_CONSTANT_ATTENUATION, atten);
+		}
 	}
 }
 
@@ -882,8 +933,12 @@ void LLLightState::setLinearAttenuation(const F32& atten)
 {
 	if (mLinearAtten != atten)
 	{
+		++gGL.mLightHash;
 		mLinearAtten = atten;
-		glLightf(GL_LIGHT0+mIndex, GL_LINEAR_ATTENUATION, atten);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightf(GL_LIGHT0+mIndex, GL_LINEAR_ATTENUATION, atten);
+		}
 	}
 }
 
@@ -891,8 +946,12 @@ void LLLightState::setQuadraticAttenuation(const F32& atten)
 {
 	if (mQuadraticAtten != atten)
 	{
+		++gGL.mLightHash;
 		mQuadraticAtten = atten;
-		glLightf(GL_LIGHT0+mIndex, GL_QUADRATIC_ATTENUATION, atten);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightf(GL_LIGHT0+mIndex, GL_QUADRATIC_ATTENUATION, atten);
+		}
 	}
 }
 
@@ -900,8 +959,12 @@ void LLLightState::setSpotExponent(const F32& exponent)
 {
 	if (mSpotExponent != exponent)
 	{
+		++gGL.mLightHash;
 		mSpotExponent = exponent;
-		glLightf(GL_LIGHT0+mIndex, GL_SPOT_EXPONENT, exponent);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightf(GL_LIGHT0+mIndex, GL_SPOT_EXPONENT, exponent);
+		}
 	}
 }
 
@@ -909,21 +972,39 @@ void LLLightState::setSpotCutoff(const F32& cutoff)
 {
 	if (mSpotCutoff != cutoff)
 	{
+		++gGL.mLightHash;
 		mSpotCutoff = cutoff;
-		glLightf(GL_LIGHT0+mIndex, GL_SPOT_CUTOFF, cutoff);
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightf(GL_LIGHT0+mIndex, GL_SPOT_CUTOFF, cutoff);
+		}
 	}
 }
 
 void LLLightState::setSpotDirection(const LLVector3& direction)
 {
 	//always set direction because modelview matrix may have changed
+	++gGL.mLightHash;
 	mSpotDirection = direction;
-	glLightfv(GL_LIGHT0+mIndex, GL_SPOT_DIRECTION, direction.mV);
+	if (!LLGLSLShader::sNoFixedFunction)
+	{
+		glLightfv(GL_LIGHT0+mIndex, GL_SPOT_DIRECTION, direction.mV);
+	}
+	else
+	{ //transform direction by current modelview matrix
+		glh::vec3f dir(direction.mV);
+
+		const glh::matrix4f& mat = gGL.getModelviewMatrix();
+		mat.mult_matrix_dir(dir);
+
+		mSpotDirection.set(dir.v);
+	}
 }
 
 LLRender::LLRender()
   : mDirty(false),
     mCount(0),
+	mQuadCycle(0),
     mMode(LLRender::TRIANGLES),
     mCurrTextureUnitIndex(0),
     mMaxAnisotropy(0.f) 
@@ -951,6 +1032,17 @@ LLRender::LLRender()
 	mCurrBlendAlphaSFactor = BF_UNDEF;
 	mCurrBlendColorDFactor = BF_UNDEF;
 	mCurrBlendAlphaDFactor = BF_UNDEF;
+
+	mMatrixMode = LLRender::MM_MODELVIEW;
+	
+	for (U32 i = 0; i < NUM_MATRIX_MODES; ++i)
+	{
+		mMatIdx[i] = 0;
+		mMatHash[i] = 0;
+		mCurMatHash[i] = 0xFFFFFFFF;
+	}
+
+	mLightHash = 0;
 }
 
 LLRender::~LLRender()
@@ -961,12 +1053,13 @@ LLRender::~LLRender()
 void LLRender::init()
 {
 	llassert_always(mBuffer.isNull()) ;
-
+	stop_glerror();
 	mBuffer = new LLVertexBuffer(immediate_mask, 0);
 	mBuffer->allocateBuffer(4096, 0, TRUE);
 	mBuffer->getVertexStrider(mVerticesp);
 	mBuffer->getTexCoord0Strider(mTexcoordsp);
 	mBuffer->getColorStrider(mColorsp);
+	stop_glerror();
 }
 
 void LLRender::shutdown()
@@ -1007,28 +1100,355 @@ void LLRender::refreshState(void)
 	mDirty = false;
 }
 
+void LLRender::syncLightState()
+{
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+
+	if (!shader)
+	{
+		return;
+	}
+
+	if (shader->mLightHash != mLightHash)
+	{
+		shader->mLightHash = mLightHash;
+
+		LLVector4 position[8];
+		LLVector3 direction[8];
+		LLVector3 attenuation[8];
+		LLVector3 diffuse[8];
+
+		for (U32 i = 0; i < 8; i++)
+		{
+			LLLightState* light = mLightState[i];
+
+			position[i] = light->mPosition;
+			direction[i] = light->mSpotDirection;
+			attenuation[i].set(light->mLinearAtten, light->mQuadraticAtten, light->mSpecular.mV[3]);
+			diffuse[i].set(light->mDiffuse.mV);
+		}
+
+		shader->uniform4fv(LLShaderMgr::LIGHT_POSITION, 8, position[0].mV);
+		shader->uniform3fv(LLShaderMgr::LIGHT_DIRECTION, 8, direction[0].mV);
+		shader->uniform3fv(LLShaderMgr::LIGHT_ATTENUATION, 8, attenuation[0].mV);
+		shader->uniform3fv(LLShaderMgr::LIGHT_DIFFUSE, 8, diffuse[0].mV);
+		shader->uniform4fv(LLShaderMgr::LIGHT_AMBIENT, 1, mAmbientLightColor.mV);
+		//HACK -- duplicate sunlight color for compatibility with drivers that can't deal with multiple shader objects referencing the same uniform
+		shader->uniform4fv(LLShaderMgr::SUNLIGHT_COLOR, 1, diffuse[0].mV);
+	}
+}
+
+void LLRender::syncMatrices()
+{
+	stop_glerror();
+
+	U32 name[] = 
+	{
+		LLShaderMgr::MODELVIEW_MATRIX,
+		LLShaderMgr::PROJECTION_MATRIX,
+		LLShaderMgr::TEXTURE_MATRIX0,
+		LLShaderMgr::TEXTURE_MATRIX1,
+		LLShaderMgr::TEXTURE_MATRIX2,
+		LLShaderMgr::TEXTURE_MATRIX3,
+	};
+
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+
+	static glh::matrix4f cached_mvp;
+	static U32 cached_mvp_mdv_hash = 0xFFFFFFFF;
+	static U32 cached_mvp_proj_hash = 0xFFFFFFFF;
+	
+	static glh::matrix4f cached_normal;
+	static U32 cached_normal_hash = 0xFFFFFFFF;
+
+	if (shader)
+	{
+		llassert(shader);
+
+		bool mvp_done = false;
+
+		U32 i = MM_MODELVIEW;
+		if (mMatHash[i] != shader->mMatHash[i])
+		{ //update modelview, normal, and MVP
+			glh::matrix4f& mat = mMatrix[i][mMatIdx[i]];
+
+			shader->uniformMatrix4fv(name[i], 1, GL_FALSE, mat.m);
+			shader->mMatHash[i] = mMatHash[i];
+
+			//update normal matrix
+			S32 loc = shader->getUniformLocation(LLShaderMgr::NORMAL_MATRIX);
+			if (loc > -1)
+			{
+				if (cached_normal_hash != mMatHash[i])
+				{
+					cached_normal = mat.inverse().transpose();
+					cached_normal_hash = mMatHash[i];
+				}
+
+				glh::matrix4f& norm = cached_normal;
+
+				F32 norm_mat[] = 
+				{
+					norm.m[0], norm.m[1], norm.m[2],
+					norm.m[4], norm.m[5], norm.m[6],
+					norm.m[8], norm.m[9], norm.m[10] 
+				};
+
+				shader->uniformMatrix3fv(LLShaderMgr::NORMAL_MATRIX, 1, GL_FALSE, norm_mat);
+			}
+
+			//update MVP matrix
+			mvp_done = true;
+			loc = shader->getUniformLocation(LLShaderMgr::MODELVIEW_PROJECTION_MATRIX);
+			if (loc > -1)
+			{
+				U32 proj = MM_PROJECTION;
+
+				if (cached_mvp_mdv_hash != mMatHash[i] || cached_mvp_proj_hash != mMatHash[MM_PROJECTION])
+				{
+					cached_mvp = mat;
+					cached_mvp.mult_left(mMatrix[proj][mMatIdx[proj]]);
+					cached_mvp_mdv_hash = mMatHash[i];
+					cached_mvp_proj_hash = mMatHash[MM_PROJECTION];
+				}
+
+				shader->uniformMatrix4fv(LLShaderMgr::MODELVIEW_PROJECTION_MATRIX, 1, GL_FALSE, cached_mvp.m);
+			}
+		}
+
+
+		i = MM_PROJECTION;
+		if (mMatHash[i] != shader->mMatHash[i])
+		{ //update projection matrix, normal, and MVP
+			glh::matrix4f& mat = mMatrix[i][mMatIdx[i]];
+
+			shader->uniformMatrix4fv(name[i], 1, GL_FALSE, mat.m);
+			shader->mMatHash[i] = mMatHash[i];
+
+			if (!mvp_done)
+			{
+				//update MVP matrix
+				S32 loc = shader->getUniformLocation(LLShaderMgr::MODELVIEW_PROJECTION_MATRIX);
+				if (loc > -1)
+				{
+					if (cached_mvp_mdv_hash != mMatHash[i] || cached_mvp_proj_hash != mMatHash[MM_PROJECTION])
+					{
+						U32 mdv = MM_MODELVIEW;
+						cached_mvp = mat;
+						cached_mvp.mult_right(mMatrix[mdv][mMatIdx[mdv]]);
+						cached_mvp_mdv_hash = mMatHash[MM_MODELVIEW];
+						cached_mvp_proj_hash = mMatHash[MM_PROJECTION];
+					}
+									
+					shader->uniformMatrix4fv(LLShaderMgr::MODELVIEW_PROJECTION_MATRIX, 1, GL_FALSE, cached_mvp.m);
+				}
+			}
+		}
+
+		for (i = MM_TEXTURE0; i < NUM_MATRIX_MODES; ++i)
+		{
+			if (mMatHash[i] != shader->mMatHash[i])
+			{
+				shader->uniformMatrix4fv(name[i], 1, GL_FALSE, mMatrix[i][mMatIdx[i]].m);
+				shader->mMatHash[i] = mMatHash[i];
+			}
+		}
+
+
+		if (shader->mFeatures.hasLighting || shader->mFeatures.calculatesLighting)
+		{ //also sync light state
+			syncLightState();
+		}
+	}
+	else if (!LLGLSLShader::sNoFixedFunction)
+	{
+		GLenum mode[] = 
+		{
+			GL_MODELVIEW,
+			GL_PROJECTION,
+			GL_TEXTURE,
+			GL_TEXTURE,
+			GL_TEXTURE,
+			GL_TEXTURE,
+		};
+
+		for (U32 i = 0; i < 2; ++i)
+		{
+			if (mMatHash[i] != mCurMatHash[i])
+			{
+				glMatrixMode(mode[i]);
+				glLoadMatrixf(mMatrix[i][mMatIdx[i]].m);
+				mCurMatHash[i] = mMatHash[i];
+			}
+		}
+
+		for (U32 i = 2; i < NUM_MATRIX_MODES; ++i)
+		{
+			if (mMatHash[i] != mCurMatHash[i])
+			{
+				gGL.getTexUnit(i-2)->activate();
+				glMatrixMode(mode[i]);
+				glLoadMatrixf(mMatrix[i][mMatIdx[i]].m);
+				mCurMatHash[i] = mMatHash[i];
+			}
+		}
+	}
+
+	stop_glerror();
+}
+
 void LLRender::translatef(const GLfloat& x, const GLfloat& y, const GLfloat& z)
 {
 	flush();
-	glTranslatef(x,y,z);
+
+	{
+		glh::matrix4f trans_mat(1,0,0,x,
+								0,1,0,y,
+								0,0,1,z,
+								0,0,0,1);
+	
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].mult_right(trans_mat);
+		mMatHash[mMatrixMode]++;
+	}
 }
 
 void LLRender::scalef(const GLfloat& x, const GLfloat& y, const GLfloat& z)
 {
 	flush();
-	glScalef(x,y,z);
+	
+	{
+		glh::matrix4f scale_mat(x,0,0,0,
+								0,y,0,0,
+								0,0,z,0,
+								0,0,0,1);
+	
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].mult_right(scale_mat);
+		mMatHash[mMatrixMode]++;
+	}
+}
+
+void LLRender::ortho(F32 left, F32 right, F32 bottom, F32 top, F32 zNear, F32 zFar)
+{
+	flush();
+
+	{
+
+		glh::matrix4f ortho_mat(2.f/(right-left),0,0,	-(right+left)/(right-left),
+								0,2.f/(top-bottom),0,	-(top+bottom)/(top-bottom),
+								0,0,-2.f/(zFar-zNear),	-(zFar+zNear)/(zFar-zNear),
+								0,0,0,1);
+	
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].mult_right(ortho_mat);
+		mMatHash[mMatrixMode]++;
+	}
+}
+
+void LLRender::rotatef(const GLfloat& a, const GLfloat& x, const GLfloat& y, const GLfloat& z)
+{
+	flush();
+
+	{
+		F32 r = a * DEG_TO_RAD;
+
+		F32 c = cosf(r);
+		F32 s = sinf(r);
+
+		F32 ic = 1.f-c;
+
+		glh::matrix4f rot_mat(x*x*ic+c,		x*y*ic-z*s,		x*z*ic+y*s,		0,
+							  x*y*ic+z*s,	y*y*ic+c,		y*z*ic-x*s,		0,
+							  x*z*ic-y*s,	y*z*ic+x*s,		z*z*ic+c,		0,
+							  0,0,0,1);
+	
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].mult_right(rot_mat);
+		mMatHash[mMatrixMode]++;
+	}
 }
 
 void LLRender::pushMatrix()
 {
 	flush();
-	glPushMatrix();
+	
+	{
+		if (mMatIdx[mMatrixMode] < LL_MATRIX_STACK_DEPTH-1)
+		{
+			mMatrix[mMatrixMode][mMatIdx[mMatrixMode]+1] = mMatrix[mMatrixMode][mMatIdx[mMatrixMode]];
+			++mMatIdx[mMatrixMode];
+		}
+		else
+		{
+			llwarns << "Matrix stack overflow." << llendl;
+		}
+	}
 }
 
 void LLRender::popMatrix()
 {
 	flush();
-	glPopMatrix();
+	{
+		if (mMatIdx[mMatrixMode] > 0)
+		{
+			--mMatIdx[mMatrixMode];
+			mMatHash[mMatrixMode]++;
+		}
+		else
+		{
+			llwarns << "Matrix stack underflow." << llendl;
+		}
+	}
+}
+
+void LLRender::loadMatrix(const GLfloat* m)
+{
+	flush();
+	{
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].set_value((GLfloat*) m);
+		mMatHash[mMatrixMode]++;
+	}
+}
+
+void LLRender::multMatrix(const GLfloat* m)
+{
+	flush();
+	{
+		glh::matrix4f mat((GLfloat*) m);
+	
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].mult_right(mat);
+		mMatHash[mMatrixMode]++;
+	}
+}
+
+void LLRender::matrixMode(U32 mode)
+{
+	if (mode == MM_TEXTURE)
+	{
+		mode = MM_TEXTURE0 + gGL.getCurrentTexUnitIndex();
+	}
+
+	llassert(mode < NUM_MATRIX_MODES);
+	mMatrixMode = mode;
+}
+
+void LLRender::loadIdentity()
+{
+	flush();
+
+	{
+		llassert_always(mMatrixMode < NUM_MATRIX_MODES) ;
+
+		mMatrix[mMatrixMode][mMatIdx[mMatrixMode]].make_identity();
+		mMatHash[mMatrixMode]++;
+	}
+}
+
+const glh::matrix4f& LLRender::getModelviewMatrix()
+{
+	return mMatrix[MM_MODELVIEW][mMatIdx[MM_MODELVIEW]];
+}
+
+const glh::matrix4f& LLRender::getProjectionMatrix()
+{
+	return mMatrix[MM_PROJECTION][mMatIdx[MM_PROJECTION]];
 }
 
 void LLRender::translateUI(F32 x, F32 y, F32 z)
@@ -1284,6 +1704,19 @@ LLLightState* LLRender::getLight(U32 index)
 	return NULL;
 }
 
+void LLRender::setAmbientLightColor(const LLColor4& color)
+{
+	if (color != mAmbientLightColor)
+	{
+		++mLightHash;
+		mAmbientLightColor = color;
+		if (!LLGLSLShader::sNoFixedFunction)
+		{
+			glLightModelfv(GL_LIGHT_MODEL_AMBIENT, color.mV);
+		}
+	}
+}
+
 bool LLRender::verifyTexUnitActive(U32 unitToVerify)
 {
 	if (mCurrTextureUnitIndex == unitToVerify)
@@ -1309,6 +1742,11 @@ void LLRender::begin(const GLuint& mode)
 {
 	if (mode != mMode)
 	{
+		if (mode == LLRender::QUADS)
+		{
+			mQuadCycle = 1;
+		}
+
 		if (mMode == LLRender::QUADS ||
 			mMode == LLRender::LINES ||
 			mMode == LLRender::TRIANGLES ||
@@ -1396,7 +1834,7 @@ void LLRender::flush()
 		
 		if (gDebugGL)
 		{
-			if (mMode == LLRender::QUADS)
+			if (mMode == LLRender::QUADS && !sGLCoreProfile)
 			{
 				if (mCount%4 != 0)
 				{
@@ -1421,12 +1859,34 @@ void LLRender::flush()
 			}
 		}
 
+		//store mCount in a local variable to avoid re-entrance (drawArrays may call flush)
+		U32 count = mCount;
+		mCount = 0;
+
+		if (mBuffer->useVBOs() && !mBuffer->isLocked())
+		{ //hack to only flush the part of the buffer that was updated (relies on stream draw using buffersubdata)
+			mBuffer->getVertexStrider(mVerticesp, 0, count);
+			mBuffer->getTexCoord0Strider(mTexcoordsp, 0, count);
+			mBuffer->getColorStrider(mColorsp, 0, count);
+		}
+		
+		mBuffer->flush();
 		mBuffer->setBuffer(immediate_mask);
-		mBuffer->drawArrays(mMode, 0, mCount);
+
+		if (mMode == LLRender::QUADS && sGLCoreProfile)
+		{
+			mBuffer->drawArrays(LLRender::TRIANGLES, 0, count);
+			mQuadCycle = 1;
+		}
+		else
+		{
+			mBuffer->drawArrays(mMode, 0, count);
+		}
+		
+		mVerticesp[0] = mVerticesp[count];
+		mTexcoordsp[0] = mTexcoordsp[count];
+		mColorsp[0] = mColorsp[count];
 		
-		mVerticesp[0] = mVerticesp[mCount];
-		mTexcoordsp[0] = mTexcoordsp[mCount];
-		mColorsp[0] = mColorsp[mCount];
 		mCount = 0;
 	}
 }
@@ -1434,6 +1894,17 @@ void LLRender::flush()
 void LLRender::vertex3f(const GLfloat& x, const GLfloat& y, const GLfloat& z)
 { 
 	//the range of mVerticesp, mColorsp and mTexcoordsp is [0, 4095]
+	if (mCount > 2048)
+	{ //break when buffer gets reasonably full to keep GL command buffers happy and avoid overflow below
+		switch (mMode)
+		{
+			case LLRender::POINTS: flush(); break;
+			case LLRender::TRIANGLES: if (mCount%3==0) flush(); break;
+			case LLRender::QUADS: if(mCount%4 == 0) flush(); break; 
+			case LLRender::LINES: if (mCount%2 == 0) flush(); break;
+		}
+	}
+			
 	if (mCount > 4094)
 	{
 	//	llwarns << "GL immediate mode overflow.  Some geometry not drawn." << llendl;
@@ -1450,10 +1921,29 @@ void LLRender::vertex3f(const GLfloat& x, const GLfloat& y, const GLfloat& z)
 		mVerticesp[mCount] = vert;
 	}
 
+	if (mMode == LLRender::QUADS && LLRender::sGLCoreProfile)
+	{
+		mQuadCycle++;
+		if (mQuadCycle == 4)
+		{ //copy two vertices so fourth quad element will add a triangle
+			mQuadCycle = 0;
+	
+			mCount++;
+			mVerticesp[mCount] = mVerticesp[mCount-3];
+			mColorsp[mCount] = mColorsp[mCount-3];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-3];
+
+			mCount++;
+			mVerticesp[mCount] = mVerticesp[mCount-2];
+			mColorsp[mCount] = mColorsp[mCount-2];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-2];
+		}
+	}
+
 	mCount++;
 	mVerticesp[mCount] = mVerticesp[mCount-1];
 	mColorsp[mCount] = mColorsp[mCount-1];
-	mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+	mTexcoordsp[mCount] = mTexcoordsp[mCount-1];	
 }
 
 void LLRender::vertexBatchPreTransformed(LLVector3* verts, S32 vert_count)
@@ -1464,13 +1954,50 @@ void LLRender::vertexBatchPreTransformed(LLVector3* verts, S32 vert_count)
 		return;
 	}
 
-	for (S32 i = 0; i < vert_count; i++)
+	if (sGLCoreProfile && mMode == LLRender::QUADS)
+	{ //quads are deprecated, convert to triangle list
+		S32 i = 0;
+		
+		while (i < vert_count)
+		{
+			//read first three
+			mVerticesp[mCount++] = verts[i++];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			mVerticesp[mCount++] = verts[i++];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			mVerticesp[mCount++] = verts[i++];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			//copy two
+			mVerticesp[mCount++] = verts[i-3];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			mVerticesp[mCount++] = verts[i-1];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+			
+			//copy last one
+			mVerticesp[mCount++] = verts[i++];
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+		}
+	}
+	else
 	{
-		mVerticesp[mCount] = verts[i];
+		for (S32 i = 0; i < vert_count; i++)
+		{
+			mVerticesp[mCount] = verts[i];
 
-		mCount++;
-		mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
-		mColorsp[mCount] = mColorsp[mCount-1];
+			mCount++;
+			mTexcoordsp[mCount] = mTexcoordsp[mCount-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+		}
 	}
 
 	mVerticesp[mCount] = mVerticesp[mCount-1];
@@ -1484,13 +2011,50 @@ void LLRender::vertexBatchPreTransformed(LLVector3* verts, LLVector2* uvs, S32 v
 		return;
 	}
 
-	for (S32 i = 0; i < vert_count; i++)
+	if (sGLCoreProfile && mMode == LLRender::QUADS)
+	{ //quads are deprecated, convert to triangle list
+		S32 i = 0;
+
+		while (i < vert_count)
+		{
+			//read first three
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount++] = uvs[i++];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount++] = uvs[i++];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount++] = uvs[i++];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			//copy last two
+			mVerticesp[mCount] = verts[i-3];
+			mTexcoordsp[mCount++] = uvs[i-3];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			mVerticesp[mCount] = verts[i-1];
+			mTexcoordsp[mCount++] = uvs[i-1];
+			mColorsp[mCount] = mColorsp[mCount-1];
+
+			//copy last one
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount++] = uvs[i++];
+			mColorsp[mCount] = mColorsp[mCount-1];
+		}
+	}
+	else
 	{
-		mVerticesp[mCount] = verts[i];
-		mTexcoordsp[mCount] = uvs[i];
+		for (S32 i = 0; i < vert_count; i++)
+		{
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount] = uvs[i];
 
-		mCount++;
-		mColorsp[mCount] = mColorsp[mCount-1];
+			mCount++;
+			mColorsp[mCount] = mColorsp[mCount-1];
+		}
 	}
 
 	mVerticesp[mCount] = mVerticesp[mCount-1];
@@ -1505,13 +2069,51 @@ void LLRender::vertexBatchPreTransformed(LLVector3* verts, LLVector2* uvs, LLCol
 		return;
 	}
 
-	for (S32 i = 0; i < vert_count; i++)
+	
+	if (sGLCoreProfile && mMode == LLRender::QUADS)
+	{ //quads are deprecated, convert to triangle list
+		S32 i = 0;
+
+		while (i < vert_count)
+		{
+			//read first three
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount] = uvs[i];
+			mColorsp[mCount++] = colors[i++];
+
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount] = uvs[i];
+			mColorsp[mCount++] = colors[i++];
+
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount] = uvs[i];
+			mColorsp[mCount++] = colors[i++];
+
+			//copy last two
+			mVerticesp[mCount] = verts[i-3];
+			mTexcoordsp[mCount] = uvs[i-3];
+			mColorsp[mCount++] = colors[i-3];
+
+			mVerticesp[mCount] = verts[i-1];
+			mTexcoordsp[mCount] = uvs[i-1];
+			mColorsp[mCount++] = colors[i-1];
+
+			//copy last one
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount] = uvs[i];
+			mColorsp[mCount++] = colors[i++];
+		}
+	}
+	else
 	{
-		mVerticesp[mCount] = verts[i];
-		mTexcoordsp[mCount] = uvs[i];
-		mColorsp[mCount] = colors[i];
+		for (S32 i = 0; i < vert_count; i++)
+		{
+			mVerticesp[mCount] = verts[i];
+			mTexcoordsp[mCount] = uvs[i];
+			mColorsp[mCount] = colors[i];
 
-		mCount++;
+			mCount++;
+		}
 	}
 
 	mVerticesp[mCount] = mVerticesp[mCount-1];
@@ -1586,6 +2188,81 @@ void LLRender::color3fv(const GLfloat* c)
 	color4f(c[0],c[1],c[2],1);
 }
 
+void LLRender::diffuseColor3f(F32 r, F32 g, F32 b)
+{
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+
+	if (shader)
+	{
+		shader->uniform4f(LLShaderMgr::DIFFUSE_COLOR, r,g,b,1.f);
+	}
+	else
+	{
+		glColor3f(r,g,b);
+	}
+}
+
+void LLRender::diffuseColor3fv(const F32* c)
+{
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+
+	if (shader)
+	{
+		shader->uniform4f(LLShaderMgr::DIFFUSE_COLOR, c[0], c[1], c[2], 1.f);
+	}
+	else
+	{
+		glColor3fv(c);
+	}
+}
+
+void LLRender::diffuseColor4f(F32 r, F32 g, F32 b, F32 a)
+{
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+
+	if (shader)
+	{
+		shader->uniform4f(LLShaderMgr::DIFFUSE_COLOR, r,g,b,a);
+	}
+	else
+	{
+		glColor4f(r,g,b,a);
+	}
+}
+
+void LLRender::diffuseColor4fv(const F32* c)
+{
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+
+	if (shader)
+	{
+		shader->uniform4fv(LLShaderMgr::DIFFUSE_COLOR, 1, c);
+	}
+	else
+	{
+		glColor4fv(c);
+	}
+}
+
+void LLRender::diffuseColor4ubv(const U8* c)
+{
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+
+	if (shader)
+	{
+		shader->uniform4f(LLShaderMgr::DIFFUSE_COLOR, c[0]/255.f, c[1]/255.f, c[2]/255.f, c[3]/255.f);
+	}
+	else
+	{
+		glColor4ubv(c);
+	}
+}
+
 void LLRender::debugTexUnits(void)
 {
 	LL_INFOS("TextureUnit") << "Active TexUnit: " << mCurrTextureUnitIndex << LL_ENDL;
diff --git a/indra/llrender/llrender.h b/indra/llrender/llrender.h
index 8f7ee30d87..fa5f7f311d 100644
--- a/indra/llrender/llrender.h
+++ b/indra/llrender/llrender.h
@@ -41,6 +41,8 @@
 #include "llstrider.h"
 #include "llpointer.h"
 #include "llglheaders.h"
+#include "llmatrix4a.h"
+#include "glh/glh_linear.h"
 
 class LLVertexBuffer;
 class LLCubeMap;
@@ -48,6 +50,8 @@ class LLImageGL;
 class LLRenderTarget;
 class LLTexture ;
 
+#define LL_MATRIX_STACK_DEPTH 32
+
 class LLTexUnit
 {
 	friend class LLRender;
@@ -235,6 +239,8 @@ public:
 	void setSpotDirection(const LLVector3& direction);
 
 protected:
+	friend class LLRender;
+
 	S32 mIndex;
 	bool mEnabled;
 	LLColor4 mDiffuse;
@@ -308,6 +314,18 @@ public:
 		BF_UNDEF
 	} eBlendFactor;
 
+	typedef enum
+	{
+		MM_MODELVIEW = 0,
+		MM_PROJECTION,
+		MM_TEXTURE0,
+		MM_TEXTURE1,
+		MM_TEXTURE2,
+		MM_TEXTURE3,
+		NUM_MATRIX_MODES,
+		MM_TEXTURE
+	} eMatrixMode;
+
 	LLRender();
 	~LLRender();
 	void init() ;
@@ -319,8 +337,21 @@ public:
 
 	void translatef(const GLfloat& x, const GLfloat& y, const GLfloat& z);
 	void scalef(const GLfloat& x, const GLfloat& y, const GLfloat& z);
+	void rotatef(const GLfloat& a, const GLfloat& x, const GLfloat& y, const GLfloat& z);
+	void ortho(F32 left, F32 right, F32 bottom, F32 top, F32 zNear, F32 zFar);
+
 	void pushMatrix();
 	void popMatrix();
+	void loadMatrix(const GLfloat* m);
+	void loadIdentity();
+	void multMatrix(const GLfloat* m);
+	void matrixMode(U32 mode);	
+
+	const glh::matrix4f& getModelviewMatrix();
+	const glh::matrix4f& getProjectionMatrix();
+
+	void syncMatrices();
+	void syncLightState();
 
 	void translateUI(F32 x, F32 y, F32 z);
 	void scaleUI(F32 x, F32 y, F32 z);
@@ -351,6 +382,12 @@ public:
 	void color3fv(const GLfloat* c);
 	void color4ubv(const GLubyte* c);
 
+	void diffuseColor3f(F32 r, F32 g, F32 b);
+	void diffuseColor3fv(const F32* c);
+	void diffuseColor4f(F32 r, F32 g, F32 b, F32 a);
+	void diffuseColor4fv(const F32* c);
+	void diffuseColor4ubv(const U8* c);
+
 	void vertexBatchPreTransformed(LLVector3* verts, S32 vert_count);
 	void vertexBatchPreTransformed(LLVector3* verts, LLVector2* uvs, S32 vert_count);
 	void vertexBatchPreTransformed(LLVector3* verts, LLVector2* uvs, LLColor4U*, S32 vert_count);
@@ -368,7 +405,8 @@ public:
 		       eBlendFactor alpha_sfactor, eBlendFactor alpha_dfactor);
 
 	LLLightState* getLight(U32 index);
-
+	void setAmbientLightColor(const LLColor4& color);
+	
 	LLTexUnit* getTexUnit(U32 index);
 
 	U32	getCurrentTexUnitIndex(void) const { return mCurrTextureUnitIndex; }
@@ -389,9 +427,21 @@ public:
 public:
 	static U32 sUICalls;
 	static U32 sUIVerts;
+	static bool sGLCoreProfile;
 	
 private:
-	bool				mDirty;
+	friend class LLLightState;
+
+	U32 mMatrixMode;
+	U32 mMatIdx[NUM_MATRIX_MODES];
+	U32 mMatHash[NUM_MATRIX_MODES];
+	glh::matrix4f mMatrix[NUM_MATRIX_MODES][LL_MATRIX_STACK_DEPTH];
+	U32 mCurMatHash[NUM_MATRIX_MODES];
+	U32 mLightHash;
+	LLColor4 mAmbientLightColor;
+	
+	bool			mDirty;
+	U32				mQuadCycle;
 	U32				mCount;
 	U32				mMode;
 	U32				mCurrTextureUnitIndex;
@@ -419,10 +469,10 @@ private:
 
 };
 
-extern F64 gGLModelView[16];
-extern F64 gGLLastModelView[16];
-extern F64 gGLLastProjection[16];
-extern F64 gGLProjection[16];
+extern F32 gGLModelView[16];
+extern F32 gGLLastModelView[16];
+extern F32 gGLLastProjection[16];
+extern F32 gGLProjection[16];
 extern S32 gGLViewport[4];
 
 extern LLRender gGL;
diff --git a/indra/llrender/llrendersphere.cpp b/indra/llrender/llrendersphere.cpp
index a5cd70445f..26bfe036e8 100644
--- a/indra/llrender/llrendersphere.cpp
+++ b/indra/llrender/llrendersphere.cpp
@@ -35,106 +35,12 @@
 
 #include "llglheaders.h"
 
-GLUquadricObj *gQuadObj2 = NULL;
 LLRenderSphere gSphere;
 
-void drawSolidSphere(GLdouble radius, GLint slices, GLint stacks);
-
-void drawSolidSphere(GLdouble radius, GLint slices, GLint stacks)
-{
-	if (!gQuadObj2)
-	{
-		gQuadObj2 = gluNewQuadric();
-		if (!gQuadObj2)
-		{
-			llwarns << "drawSolidSphere couldn't allocate quadric" << llendl;
-			return;
-		}
-	}
-
-	gluQuadricDrawStyle(gQuadObj2, GLU_FILL);
-	gluQuadricNormals(gQuadObj2, GLU_SMOOTH);
-	// If we ever changed/used the texture or orientation state
-	// of quadObj, we'd need to change it to the defaults here
-	// with gluQuadricTexture and/or gluQuadricOrientation.
-	gluQuadricTexture(gQuadObj2, GL_TRUE);
-	gluSphere(gQuadObj2, radius, slices, stacks);
-}
-
-
-// A couple thoughts on sphere drawing:
-// 1) You need more slices than stacks, but little less than 2:1
-// 2) At low LOD, setting stacks to an odd number avoids a "band" around the equator, making things look smoother
-void LLRenderSphere::prerender()
-{
-	//  Create a series of display lists for different LODs
-	mDList[0] = glGenLists(1);
-	glNewList(mDList[0], GL_COMPILE);
-	drawSolidSphere(1.0, 30, 20);
-	glEndList();
-
-	mDList[1] = glGenLists(1);
-	glNewList(mDList[1], GL_COMPILE);
-	drawSolidSphere(1.0, 20, 15);
-	glEndList();
-
-	mDList[2] = glGenLists(1);
-	glNewList(mDList[2], GL_COMPILE);
-	drawSolidSphere(1.0, 12, 8);
-	glEndList();
-
-	mDList[3] = glGenLists(1);
-	glNewList(mDList[3], GL_COMPILE);
-	drawSolidSphere(1.0, 8, 5);
-	glEndList();
-}
-
-void LLRenderSphere::cleanupGL()
-{
-	for (S32 detail = 0; detail < 4; detail++)
-	{
-		glDeleteLists(mDList[detail], 1);
-		mDList[detail] = 0;
-	}
-	
-	if (gQuadObj2)
-	{
-		gluDeleteQuadric(gQuadObj2);
-		gQuadObj2 = NULL;
-	}
-}
-
-// Constants here are empirically derived from my eyeballs, JNC
-//
-// The toughest adjustment is the cutoff for the lowest LOD
-// Maybe we should have more LODs at the low end?
-void LLRenderSphere::render(F32 pixel_area)
-{
-	S32 level_of_detail;
-
-	if (pixel_area > 10000.f)
-	{
-		level_of_detail = 0;
-	}
-	else if (pixel_area > 800.f)
-	{
-		level_of_detail = 1;
-	}
-	else if (pixel_area > 100.f)
-	{
-		level_of_detail = 2;
-	}
-	else
-	{
-		level_of_detail = 3;
-	}
-	glCallList(mDList[level_of_detail]);
-}
-
-
 void LLRenderSphere::render()
 {
-	glCallList(mDList[0]);
+	renderGGL();
+	gGL.flush();
 }
 
 inline LLVector3 polar_to_cart(F32 latitude, F32 longitude)
diff --git a/indra/llrender/llrendersphere.h b/indra/llrender/llrendersphere.h
index 96a6bec80c..f8e9e86e7f 100644
--- a/indra/llrender/llrendersphere.h
+++ b/indra/llrender/llrendersphere.h
@@ -40,11 +40,6 @@ void lat2xyz(LLVector3 * result, F32 lat, F32 lon);			// utility routine
 class LLRenderSphere  
 {
 public:
-	LLGLuint	mDList[5];
-
-	void prerender();
-	void cleanupGL();
-	void render(F32 pixel_area);		// of a box of size 1.0 at that position
 	void render();						// render at highest LOD
 	void renderGGL();                   // render using LLRender
 
diff --git a/indra/llrender/llrendertarget.cpp b/indra/llrender/llrendertarget.cpp
index 8c0d3592df..ef2a7395da 100644
--- a/indra/llrender/llrendertarget.cpp
+++ b/indra/llrender/llrendertarget.cpp
@@ -31,8 +31,7 @@
 #include "llgl.h"
 
 LLRenderTarget* LLRenderTarget::sBoundTarget = NULL;
-
-
+U32 LLRenderTarget::sBytesAllocated = 0;
 
 void check_framebuffer_status()
 {
@@ -62,8 +61,7 @@ LLRenderTarget::LLRenderTarget() :
 	mStencil(0),
 	mUseDepth(false),
 	mRenderDepth(false),
-	mUsage(LLTexUnit::TT_TEXTURE),
-	mSamples(0)
+	mUsage(LLTexUnit::TT_TEXTURE)
 {
 }
 
@@ -84,20 +82,6 @@ bool LLRenderTarget::allocate(U32 resx, U32 resy, U32 color_fmt, bool depth, boo
 	mStencil = stencil;
 	mUsage = usage;
 	mUseDepth = depth;
-	mSamples = samples;
-
-	mSamples = gGLManager.getNumFBOFSAASamples(mSamples);
-	
-	if (mSamples > 1 && gGLManager.mHasTextureMultisample)
-	{
-		mUsage = LLTexUnit::TT_MULTISAMPLE_TEXTURE;
-		//no support for multisampled stencil targets yet
-		mStencil = false;
-	}
-	else
-	{
-		mSamples = 0;
-	}
 
 	if ((sUseFBO || use_fbo) && gGLManager.mHasFramebufferObject)
 	{
@@ -157,21 +141,6 @@ bool LLRenderTarget::addColorAttachment(U32 color_fmt)
 	stop_glerror();
 
 
-#ifdef GL_ARB_texture_multisample
-	if (mSamples > 1)
-	{
-		clear_glerror();
-		glTexImage2DMultisample(LLTexUnit::getInternalType(mUsage), mSamples, color_fmt, mResX, mResY, GL_TRUE);
-		if (glGetError() != GL_NO_ERROR)
-		{
-			llwarns << "Could not allocate multisample color buffer for render target." << llendl;
-			return false;
-		}
-	}
-	else
-#else
-	llassert_always(mSamples <= 1);
-#endif
 	{
 		clear_glerror();
 		LLImageGL::setManualImage(LLTexUnit::getInternalType(mUsage), 0, color_fmt, mResX, mResY, GL_RGBA, GL_UNSIGNED_BYTE, NULL);
@@ -182,32 +151,32 @@ bool LLRenderTarget::addColorAttachment(U32 color_fmt)
 		}
 	}
 	
+	sBytesAllocated += mResX*mResY*4;
+
 	stop_glerror();
 
-	if (mSamples == 0)
-	{ 
-		if (offset == 0)
-		{ //use bilinear filtering on single texture render targets that aren't multisampled
-			gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_BILINEAR);
-			stop_glerror();
-		}
-		else
-		{ //don't filter data attachments
-			gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_POINT);
-			stop_glerror();
-		}
+	
+	if (offset == 0)
+	{ //use bilinear filtering on single texture render targets that aren't multisampled
+		gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_BILINEAR);
+		stop_glerror();
+	}
+	else
+	{ //don't filter data attachments
+		gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_POINT);
+		stop_glerror();
+	}
 
-		if (mUsage != LLTexUnit::TT_RECT_TEXTURE)
-		{
-			gGL.getTexUnit(0)->setTextureAddressMode(LLTexUnit::TAM_MIRROR);
-			stop_glerror();
-		}
-		else
-		{
-			// ATI doesn't support mirrored repeat for rectangular textures.
-			gGL.getTexUnit(0)->setTextureAddressMode(LLTexUnit::TAM_CLAMP);
-			stop_glerror();
-		}
+	if (mUsage != LLTexUnit::TT_RECT_TEXTURE)
+	{
+		gGL.getTexUnit(0)->setTextureAddressMode(LLTexUnit::TAM_MIRROR);
+		stop_glerror();
+	}
+	else
+	{
+		// ATI doesn't support mirrored repeat for rectangular textures.
+		gGL.getTexUnit(0)->setTextureAddressMode(LLTexUnit::TAM_CLAMP);
+		stop_glerror();
 	}
 		
 	if (mFBO)
@@ -250,26 +219,16 @@ bool LLRenderTarget::allocateDepth()
 	{
 		LLImageGL::generateTextures(1, &mDepth);
 		gGL.getTexUnit(0)->bindManual(mUsage, mDepth);
-		if (mSamples == 0)
-		{
-			U32 internal_type = LLTexUnit::getInternalType(mUsage);
-			gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_POINT);
-			stop_glerror();
-			clear_glerror();
-			LLImageGL::setManualImage(internal_type, 0, GL_DEPTH_COMPONENT32, mResX, mResY, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, NULL);
-		}
-#ifdef GL_ARB_texture_multisample
-		else
-		{
-			stop_glerror();
-			clear_glerror();
-			glTexImage2DMultisample(LLTexUnit::getInternalType(mUsage), mSamples, GL_DEPTH_COMPONENT32, mResX, mResY, GL_TRUE);
-		}
-#else
-		llassert_always(mSamples <= 1);
-#endif
+		
+		U32 internal_type = LLTexUnit::getInternalType(mUsage);
+		stop_glerror();
+		clear_glerror();
+		LLImageGL::setManualImage(internal_type, 0, GL_DEPTH_COMPONENT24, mResX, mResY, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, NULL);
+		gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_POINT);
 	}
 
+	sBytesAllocated += mResX*mResY*4;
+
 	if (glGetError() != GL_NO_ERROR)
 	{
 		llwarns << "Unable to allocate depth buffer for render target." << llendl;
@@ -339,14 +298,16 @@ void LLRenderTarget::release()
 			stop_glerror();
 		}
 		mDepth = 0;
+
+		sBytesAllocated -= mResX*mResY*4;
 	}
 	else if (mUseDepth && mFBO)
 	{ //detach shared depth buffer
 		glBindFramebuffer(GL_FRAMEBUFFER, mFBO);
 		if (mStencil)
 		{ //attached as a renderbuffer
-			glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, 0);
 			glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_RENDERBUFFER, 0);
+			glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, 0);
 			mStencil = false;
 		}
 		else
@@ -364,6 +325,7 @@ void LLRenderTarget::release()
 
 	if (mTex.size() > 0)
 	{
+		sBytesAllocated -= mResX*mResY*4*mTex.size();
 		LLImageGL::deleteTextures(mTex.size(), &mTex[0], true);
 		mTex.clear();
 	}
@@ -495,7 +457,8 @@ void LLRenderTarget::copyContents(LLRenderTarget& source, S32 srcX0, S32 srcY0,
 	gGL.flush();
 	if (!source.mFBO || !mFBO)
 	{
-		llerrs << "Cannot copy framebuffer contents for non FBO render targets." << llendl;
+		llwarns << "Cannot copy framebuffer contents for non FBO render targets." << llendl;
+		return;
 	}
 
 	
diff --git a/indra/llrender/llrendertarget.h b/indra/llrender/llrendertarget.h
index dea1de12d8..2735ab21c5 100644
--- a/indra/llrender/llrendertarget.h
+++ b/indra/llrender/llrendertarget.h
@@ -64,6 +64,7 @@ class LLRenderTarget
 public:
 	//whether or not to use FBO implementation
 	static bool sUseFBO; 
+	static U32 sBytesAllocated;
 
 	LLRenderTarget();
 	~LLRenderTarget();
@@ -147,7 +148,6 @@ protected:
 	bool mUseDepth;
 	bool mRenderDepth;
 	LLTexUnit::eTextureType mUsage;
-	U32 mSamples;
 	
 	static LLRenderTarget* sBoundTarget;
 };
diff --git a/indra/llrender/llshadermgr.cpp b/indra/llrender/llshadermgr.cpp
index 986c1f2774..7d384450e6 100644
--- a/indra/llrender/llshadermgr.cpp
+++ b/indra/llrender/llshadermgr.cpp
@@ -81,29 +81,42 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 	// NOTE order of shader object attaching is VERY IMPORTANT!!!
 	if (features->calculatesAtmospherics)
 	{
-		if (!shader->attachObject("windlight/atmosphericsVarsV.glsl"))
+		if (features->hasWaterFog)
+		{
+			if (!shader->attachObject("windlight/atmosphericsVarsWaterV.glsl"))
+			{
+				return FALSE;
+			}
+		}
+		else if (!shader->attachObject("windlight/atmosphericsVarsV.glsl"))
 		{
 			return FALSE;
 		}
 	}
 
-	if (features->calculatesLighting)
+	if (features->calculatesLighting || features->atmosphericHelpers)
 	{
 		if (!shader->attachObject("windlight/atmosphericsHelpersV.glsl"))
 		{
 			return FALSE;
 		}
+	}
 		
+	if (features->calculatesLighting)
+	{
 		if (features->isSpecular)
 		{
 			if (!shader->attachObject("lighting/lightFuncSpecularV.glsl"))
 			{
 				return FALSE;
 			}
-			
-			if (!shader->attachObject("lighting/sumLightsSpecularV.glsl"))
+		
+			if (!features->isAlphaLighting)
 			{
-				return FALSE;
+				if (!shader->attachObject("lighting/sumLightsSpecularV.glsl"))
+				{
+					return FALSE;
+				}
 			}
 			
 			if (!shader->attachObject("lighting/lightSpecularV.glsl"))
@@ -118,9 +131,12 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 				return FALSE;
 			}
 			
-			if (!shader->attachObject("lighting/sumLightsV.glsl"))
+			if (!features->isAlphaLighting)
 			{
-				return FALSE;
+				if (!shader->attachObject("lighting/sumLightsV.glsl"))
+				{
+					return FALSE;
+				}
 			}
 			
 			if (!shader->attachObject("lighting/lightV.glsl"))
@@ -161,7 +177,14 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 
 	if(features->calculatesAtmospherics)
 	{
-		if (!shader->attachObject("windlight/atmosphericsVarsF.glsl"))
+		if (features->hasWaterFog)
+		{
+			if (!shader->attachObject("windlight/atmosphericsVarsWaterF.glsl"))
+			{
+				return FALSE;
+			}
+		}
+		else if (!shader->attachObject("windlight/atmosphericsVarsF.glsl"))
 		{
 			return FALSE;
 		}
@@ -241,7 +264,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 						return FALSE;
 					}
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 		
@@ -280,9 +303,9 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 						return FALSE;
 					}
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
-		}		
+		}
 	}
 	
 	// NOTE order of shader object attaching is VERY IMPORTANT!!!
@@ -304,7 +327,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 				{
 					return FALSE;
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 		else if (features->hasWaterFog)
@@ -336,7 +359,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 				{
 					return FALSE;
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 		
@@ -355,7 +378,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 				{
 					return FALSE;
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 		
@@ -395,7 +418,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 						return FALSE;
 					}
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 	}
@@ -419,7 +442,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 				{
 					return FALSE;
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 		
@@ -438,10 +461,26 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 				{
 					return FALSE;
 				}
-				shader->mFeatures.mIndexedTextureChannels = gGLManager.mNumTextureImageUnits-1;
+				shader->mFeatures.mIndexedTextureChannels = llmax(LLGLSLShader::sIndexedTextureChannels-1, 1);
 			}
 		}
 	}
+
+	if (features->mIndexedTextureChannels <= 1)
+	{
+		if (!shader->attachObject("objects/nonindexedTextureV.glsl"))
+		{
+			return FALSE;
+		}
+	}
+	else
+	{
+		if (!shader->attachObject("objects/indexedTextureV.glsl"))
+		{
+			return FALSE;
+		}
+	}
+
 	return TRUE;
 }
 
@@ -477,7 +516,7 @@ void LLShaderMgr::dumpObjectLog(GLhandleARB ret, BOOL warns)
 		}
 		else
 		{
-			LL_DEBUGS("ShaderLoading") << log << LL_ENDL;
+			LL_INFOS("ShaderLoading") << log << LL_ENDL;
 		}
 	}
  }
@@ -531,23 +570,85 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 	}
 
 	//we can't have any lines longer than 1024 characters 
-	//or any shaders longer than 1024 lines... deal - DaveP
+	//or any shaders longer than 4096 lines... deal - DaveP
 	GLcharARB buff[1024];
-	GLcharARB* text[1024];
+	GLcharARB* text[4096];
 	GLuint count = 0;
 
-	if (gGLManager.mGLVersion < 2.1f)
-	{
-		text[count++] = strdup("#version 110\n");
-	}
-	else if (gGLManager.mGLVersion < 3.f)
+	S32 major_version = gGLManager.mGLSLVersionMajor;
+	S32 minor_version = gGLManager.mGLSLVersionMinor;
+	
+	if (major_version == 1 && minor_version < 30)
 	{
-		//set version to 1.20
-		text[count++] = strdup("#version 120\n");
+		if (minor_version < 10)
+		{
+			//should NEVER get here -- if major version is 1 and minor version is less than 10, 
+			// viewer should never attempt to use shaders, continuing will result in undefined behavior
+			llerrs << "Unsupported GLSL Version." << llendl;
+		}
+
+		if (minor_version <= 19)
+		{
+			text[count++] = strdup("#version 110\n");
+			text[count++] = strdup("#define ATTRIBUTE attribute\n");
+			text[count++] = strdup("#define VARYING varying\n");
+			text[count++] = strdup("#define VARYING_FLAT varying\n");
+		}
+		else if (minor_version <= 29)
+		{
+			//set version to 1.20
+			text[count++] = strdup("#version 120\n");
+			text[count++] = strdup("#define FXAA_GLSL_120 1\n");
+			text[count++] = strdup("#define FXAA_FAST_PIXEL_OFFSET 0\n");
+			text[count++] = strdup("#define ATTRIBUTE attribute\n");
+			text[count++] = strdup("#define VARYING varying\n");
+			text[count++] = strdup("#define VARYING_FLAT varying\n");
+		}
 	}
 	else
-	{  //set version to 1.30
-		text[count++] = strdup("#version 130\n");
+	{  
+		if (major_version < 4)
+		{
+			//set version to 1.30
+			text[count++] = strdup("#version 130\n");
+
+			//some implementations of GLSL 1.30 require integer precision be explicitly declared
+			text[count++] = strdup("precision mediump int;\n");
+			text[count++] = strdup("precision highp float;\n");
+		}
+		else
+		{ //set version to 400
+			text[count++] = strdup("#version 400\n");
+		}
+
+		text[count++] = strdup("#define DEFINE_GL_FRAGCOLOR 1\n");
+		text[count++] = strdup("#define FXAA_GLSL_130 1\n");
+
+		text[count++] = strdup("#define ATTRIBUTE in\n");
+
+		if (type == GL_VERTEX_SHADER_ARB)
+		{ //"varying" state is "out" in a vertex program, "in" in a fragment program 
+			// ("varying" is deprecated after version 1.20)
+			text[count++] = strdup("#define VARYING out\n");
+			text[count++] = strdup("#define VARYING_FLAT flat out\n");
+		}
+		else
+		{
+			text[count++] = strdup("#define VARYING in\n");
+			text[count++] = strdup("#define VARYING_FLAT flat in\n");
+		}
+
+		//backwards compatibility with legacy texture lookup syntax
+		text[count++] = strdup("#define texture2D texture\n");
+		text[count++] = strdup("#define textureCube texture\n");
+		text[count++] = strdup("#define texture2DLod textureLod\n");
+		text[count++] = strdup("#define	shadow2D(a,b) vec2(texture(a,b))\n");
+
+		if (major_version > 1 || minor_version >= 40)
+		{ //GLSL 1.40 replaces texture2DRect et al with texture
+			text[count++] = strdup("#define texture2DRect texture\n");
+			text[count++] = strdup("#define shadow2DRect(a,b) vec2(texture(a,b))\n");
+		}
 	}
 
 	//copy preprocessor definitions into buffer
@@ -571,22 +672,24 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 		.
 		uniform sampler2D texN;
 		
-		varying float vary_texture_index;
+		VARYING_FLAT ivec4 vary_texture_index;
+
+		vec4 ret = vec4(1,0,1,1);
 
 		vec4 diffuseLookup(vec2 texcoord)
 		{
-			switch (int(vary_texture_index+0.25))
+			switch (vary_texture_index.r))
 			{
-				case 0: return texture2D(tex0, texcoord);
-				case 1: return texture2D(tex1, texcoord);
-				case 2: return texture2D(tex2, texcoord);
+				case 0: ret = texture2D(tex0, texcoord); break;
+				case 1: ret = texture2D(tex1, texcoord); break;
+				case 2: ret = texture2D(tex2, texcoord); break;
 				.
 				.
 				.
-				case N: return texture2D(texN, texcoord);
+				case N: return texture2D(texN, texcoord); break;
 			}
 
-			return vec4(0,0,0,0);
+			return ret;
 		}
 		*/
 
@@ -597,7 +700,11 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 			text[count++] = strdup(decl.c_str());
 		}
 
-		text[count++] = strdup("varying float vary_texture_index;\n");
+		if (texture_index_channels > 1)
+		{
+			text[count++] = strdup("VARYING_FLAT ivec4 vary_texture_index;\n");
+		}
+
 		text[count++] = strdup("vec4 diffuseLookup(vec2 texcoord)\n");
 		text[count++] = strdup("{\n");
 		
@@ -607,49 +714,32 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 			text[count++] = strdup("return texture2D(tex0, texcoord);\n");
 			text[count++] = strdup("}\n");
 		}
-		else if (gGLManager.mGLVersion >= 3.f)
-		{ 
-			text[count++] = strdup("\tswitch (int(vary_texture_index+0.25))\n");
+		else if (major_version > 1 || minor_version >= 30)
+		{  //switches are supported in GLSL 1.30 and later
+			text[count++] = strdup("\tvec4 ret = vec4(1,0,1,1);\n");
+			text[count++] = strdup("\tswitch (vary_texture_index.r)\n");
 			text[count++] = strdup("\t{\n");
 		
 			//switch body
 			for (S32 i = 0; i < texture_index_channels; ++i)
 			{
-				std::string case_str = llformat("\t\tcase %d: return texture2D(tex%d, texcoord);\n", i, i);
+				std::string case_str = llformat("\t\tcase %d: ret = texture2D(tex%d, texcoord); break;\n", i, i);
 				text[count++] = strdup(case_str.c_str());
 			}
 
 			text[count++] = strdup("\t}\n");
-			text[count++] = strdup("\treturn vec4(0,0,0,0);\n");
+			text[count++] = strdup("\treturn ret;\n");
 			text[count++] = strdup("}\n");
 		}
 		else
-		{
-			//switches aren't supported, make block that looks like:
-			/*
-				int ti = int(vary_texture_index+0.25);
-				if (ti == 0) return texture2D(tex0, texcoord);
-				if (ti == 1) return texture2D(tex1, texcoord);
-				.
-				.
-				.
-				if (ti == N) return texture2D(texN, texcoord);
-			*/
-				
-			text[count++] = strdup("int ti = int(vary_texture_index+0.25);\n");
-			for (S32 i = 0; i < texture_index_channels; ++i)
-			{
-				std::string if_str = llformat("if (ti == %d) return texture2D(tex%d, texcoord);\n", i, i);
-				text[count++] = strdup(if_str.c_str());
-			}
-
-			text[count++] = strdup("\treturn vec4(0,0,0,0);\n");
-			text[count++] = strdup("}\n");
-		}			
+		{ //should never get here.  Indexed texture rendering requires GLSL 1.30 or later 
+			// (for passing integers between vertex and fragment shaders)
+			llerrs << "Indexed texture rendering requires GLSL 1.30 or later." << llendl;
+		}
 	}
 
 	//copy file into memory
-	while( fgets((char *)buff, 1024, file) != NULL && count < LL_ARRAY_SIZE(buff) ) 
+	while( fgets((char *)buff, 1024, file) != NULL && count < LL_ARRAY_SIZE(text) ) 
 	{
 		text[count++] = (GLcharARB *)strdup((char *)buff); 
 	}
@@ -704,14 +794,24 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 				LL_WARNS("ShaderLoading") << "GLSL Compilation Error: (" << error << ") in " << filename << LL_ENDL;
 				dumpObjectLog(ret);
 
+#if LL_WINDOWS
 				std::stringstream ostr;
 				//dump shader source for debugging
 				for (GLuint i = 0; i < count; i++)
 				{
 					ostr << i << ": " << text[i];
+
+					if (i % 128 == 0)
+					{ //dump every 128 lines
+
+						LL_WARNS("ShaderLoading") << "\n" << ostr.str() << llendl;
+						ostr = std::stringstream();
+					}
+
 				}
 
 				LL_WARNS("ShaderLoading") << "\n" << ostr.str() << llendl;
+#endif // LL_WINDOWS
 
 				ret = 0;
 			}
@@ -760,28 +860,42 @@ BOOL LLShaderMgr::linkProgramObject(GLhandleARB obj, BOOL suppress_errors)
 		LL_WARNS("ShaderLoading") << "GLSL Linker Error:" << LL_ENDL;
 	}
 
-// NOTE: Removing LL_DARWIN block as it doesn't seem to actually give the correct answer, 
-// but want it for reference once I move it.
-#if 0
-	// Force an evaluation of the gl state so the driver can tell if the shader will run in hardware or software
-	// per Apple's suggestion   
-	glBegin(gGL.mMode);
-	glEnd();
+#if LL_DARWIN
 
-	// Query whether the shader can or cannot run in hardware
-	// http://developer.apple.com/qa/qa2007/qa1502.html
-	long vertexGPUProcessing;
-	CGLContextObj ctx = CGLGetCurrentContext();
-	CGLGetParameter (ctx, kCGLCPGPUVertexProcessing, &vertexGPUProcessing);	
-	long fragmentGPUProcessing;
-	CGLGetParameter (ctx, kCGLCPGPUFragmentProcessing, &fragmentGPUProcessing);
-	if (!fragmentGPUProcessing || !vertexGPUProcessing)
+	// For some reason this absolutely kills the frame rate when VBO's are enabled
+	if (0)
 	{
-		LL_WARNS("ShaderLoading") << "GLSL Linker: Running in Software:" << LL_ENDL;
-		success = GL_FALSE;
-		suppress_errors = FALSE;		
+		// Force an evaluation of the gl state so the driver can tell if the shader will run in hardware or software
+		// per Apple's suggestion
+		LLGLSLShader::sNoFixedFunction = false;
+		
+		glUseProgramObjectARB(obj);
+
+		gGL.begin(LLRender::TRIANGLES);
+		gGL.vertex3f(0.0f, 0.0f, 0.0f);
+		gGL.vertex3f(0.0f, 0.0f, 0.0f);
+		gGL.vertex3f(0.0f, 0.0f, 0.0f);
+		gGL.end();
+		gGL.flush();
+		
+		glUseProgramObjectARB(0);
+		
+		LLGLSLShader::sNoFixedFunction = true;
+
+		// Query whether the shader can or cannot run in hardware
+		// http://developer.apple.com/qa/qa2007/qa1502.html
+		GLint vertexGPUProcessing, fragmentGPUProcessing;
+		CGLContextObj ctx = CGLGetCurrentContext();
+		CGLGetParameter(ctx, kCGLCPGPUVertexProcessing, &vertexGPUProcessing);	
+		CGLGetParameter(ctx, kCGLCPGPUFragmentProcessing, &fragmentGPUProcessing);
+		if (!fragmentGPUProcessing || !vertexGPUProcessing)
+		{
+			LL_WARNS("ShaderLoading") << "GLSL Linker: Running in Software:" << LL_ENDL;
+			success = GL_FALSE;
+			suppress_errors = FALSE;		
+		}
 	}
-	
+
 #else
 	std::string log = get_object_log(obj);
 	LLStringUtil::toLower(log);
@@ -819,3 +933,183 @@ BOOL LLShaderMgr::validateProgramObject(GLhandleARB obj)
 	return success;
 }
 
+//virtual
+void LLShaderMgr::initAttribsAndUniforms()
+{
+	//MUST match order of enum in LLVertexBuffer.h
+	mReservedAttribs.push_back("position");
+	mReservedAttribs.push_back("normal");
+	mReservedAttribs.push_back("texcoord0");
+	mReservedAttribs.push_back("texcoord1");
+	mReservedAttribs.push_back("texcoord2");
+	mReservedAttribs.push_back("texcoord3");
+	mReservedAttribs.push_back("diffuse_color");
+	mReservedAttribs.push_back("emissive");
+	mReservedAttribs.push_back("binormal");
+	mReservedAttribs.push_back("weight");
+	mReservedAttribs.push_back("weight4");
+	mReservedAttribs.push_back("clothing");
+	mReservedAttribs.push_back("texture_index");
+	
+	//matrix state
+	mReservedUniforms.push_back("modelview_matrix");
+	mReservedUniforms.push_back("projection_matrix");
+	mReservedUniforms.push_back("inv_proj");
+	mReservedUniforms.push_back("modelview_projection_matrix");
+	mReservedUniforms.push_back("normal_matrix");
+	mReservedUniforms.push_back("texture_matrix0");
+	mReservedUniforms.push_back("texture_matrix1");
+	mReservedUniforms.push_back("texture_matrix2");
+	mReservedUniforms.push_back("texture_matrix3");
+	llassert(mReservedUniforms.size() == LLShaderMgr::TEXTURE_MATRIX3+1);
+
+	mReservedUniforms.push_back("viewport");
+
+	mReservedUniforms.push_back("light_position");
+	mReservedUniforms.push_back("light_direction");
+	mReservedUniforms.push_back("light_attenuation");
+	mReservedUniforms.push_back("light_diffuse");
+	mReservedUniforms.push_back("light_ambient");
+	mReservedUniforms.push_back("light_count");
+	mReservedUniforms.push_back("light");
+	mReservedUniforms.push_back("light_col");
+	mReservedUniforms.push_back("far_z");
+
+	llassert(mReservedUniforms.size() == LLShaderMgr::MULTI_LIGHT_FAR_Z+1);
+
+
+	mReservedUniforms.push_back("proj_mat");
+	mReservedUniforms.push_back("proj_near");
+	mReservedUniforms.push_back("proj_p");
+	mReservedUniforms.push_back("proj_n");
+	mReservedUniforms.push_back("proj_origin");
+	mReservedUniforms.push_back("proj_range");
+	mReservedUniforms.push_back("proj_ambiance");
+	mReservedUniforms.push_back("proj_shadow_idx");
+	mReservedUniforms.push_back("shadow_fade");
+	mReservedUniforms.push_back("proj_focus");
+	mReservedUniforms.push_back("proj_lod");
+	mReservedUniforms.push_back("proj_ambient_lod");
+
+	llassert(mReservedUniforms.size() == LLShaderMgr::PROJECTOR_AMBIENT_LOD+1);
+
+	mReservedUniforms.push_back("color");
+		
+	mReservedUniforms.push_back("diffuseMap");
+	mReservedUniforms.push_back("specularMap");
+	mReservedUniforms.push_back("bumpMap");
+	mReservedUniforms.push_back("environmentMap");
+	mReservedUniforms.push_back("cloude_noise_texture");
+	mReservedUniforms.push_back("fullbright");
+	mReservedUniforms.push_back("lightnorm");
+	mReservedUniforms.push_back("sunlight_color_copy");
+	mReservedUniforms.push_back("ambient");
+	mReservedUniforms.push_back("blue_horizon");
+	mReservedUniforms.push_back("blue_density");
+	mReservedUniforms.push_back("haze_horizon");
+	mReservedUniforms.push_back("haze_density");
+	mReservedUniforms.push_back("cloud_shadow");
+	mReservedUniforms.push_back("density_multiplier");
+	mReservedUniforms.push_back("distance_multiplier");
+	mReservedUniforms.push_back("max_y");
+	mReservedUniforms.push_back("glow");
+	mReservedUniforms.push_back("cloud_color");
+	mReservedUniforms.push_back("cloud_pos_density1");
+	mReservedUniforms.push_back("cloud_pos_density2");
+	mReservedUniforms.push_back("cloud_scale");
+	mReservedUniforms.push_back("gamma");
+	mReservedUniforms.push_back("scene_light_strength");
+
+	llassert(mReservedUniforms.size() == LLShaderMgr::SCENE_LIGHT_STRENGTH+1);
+
+	mReservedUniforms.push_back("center");
+	mReservedUniforms.push_back("size");
+	mReservedUniforms.push_back("falloff");
+
+
+	mReservedUniforms.push_back("minLuminance");
+	mReservedUniforms.push_back("maxExtractAlpha");
+	mReservedUniforms.push_back("lumWeights");
+	mReservedUniforms.push_back("warmthWeights");
+	mReservedUniforms.push_back("warmthAmount");
+	mReservedUniforms.push_back("glowStrength");
+	mReservedUniforms.push_back("glowDelta");
+
+	llassert(mReservedUniforms.size() == LLShaderMgr::GLOW_DELTA+1);
+
+
+	mReservedUniforms.push_back("minimum_alpha");
+
+	mReservedUniforms.push_back("shadow_matrix");
+	mReservedUniforms.push_back("env_mat");
+	mReservedUniforms.push_back("shadow_clip");
+	mReservedUniforms.push_back("sun_wash");
+	mReservedUniforms.push_back("shadow_noise");
+	mReservedUniforms.push_back("blur_size");
+	mReservedUniforms.push_back("ssao_radius");
+	mReservedUniforms.push_back("ssao_max_radius");
+	mReservedUniforms.push_back("ssao_factor");
+	mReservedUniforms.push_back("ssao_factor_inv");
+	mReservedUniforms.push_back("ssao_effect_mat");
+	mReservedUniforms.push_back("screen_res");
+	mReservedUniforms.push_back("near_clip");
+	mReservedUniforms.push_back("shadow_offset");
+	mReservedUniforms.push_back("shadow_bias");
+	mReservedUniforms.push_back("spot_shadow_bias");
+	mReservedUniforms.push_back("spot_shadow_offset");
+	mReservedUniforms.push_back("sun_dir");
+	mReservedUniforms.push_back("shadow_res");
+	mReservedUniforms.push_back("proj_shadow_res");
+	mReservedUniforms.push_back("depth_cutoff");
+	mReservedUniforms.push_back("norm_cutoff");
+	
+	llassert(mReservedUniforms.size() == LLShaderMgr::DEFERRED_NORM_CUTOFF+1);
+
+	mReservedUniforms.push_back("tc_scale");
+	mReservedUniforms.push_back("rcp_screen_res");
+	mReservedUniforms.push_back("rcp_frame_opt");
+	mReservedUniforms.push_back("rcp_frame_opt2");
+	
+	mReservedUniforms.push_back("focal_distance");
+	mReservedUniforms.push_back("blur_constant");
+	mReservedUniforms.push_back("tan_pixel_angle");
+	mReservedUniforms.push_back("magnification");
+	mReservedUniforms.push_back("max_cof");
+	mReservedUniforms.push_back("res_scale");
+	mReservedUniforms.push_back("dof_width");
+	mReservedUniforms.push_back("dof_height");
+
+	mReservedUniforms.push_back("depthMap");
+	mReservedUniforms.push_back("shadowMap0");
+	mReservedUniforms.push_back("shadowMap1");
+	mReservedUniforms.push_back("shadowMap2");
+	mReservedUniforms.push_back("shadowMap3");
+	mReservedUniforms.push_back("shadowMap4");
+	mReservedUniforms.push_back("shadowMap5");
+
+	llassert(mReservedUniforms.size() == LLShaderMgr::DEFERRED_SHADOW5+1);
+
+	mReservedUniforms.push_back("normalMap");
+	mReservedUniforms.push_back("positionMap");
+	mReservedUniforms.push_back("diffuseRect");
+	mReservedUniforms.push_back("specularRect");
+	mReservedUniforms.push_back("noiseMap");
+	mReservedUniforms.push_back("lightFunc");
+	mReservedUniforms.push_back("lightMap");
+	mReservedUniforms.push_back("bloomMap");
+	mReservedUniforms.push_back("projectionMap");
+
+	llassert(mReservedUniforms.size() == END_RESERVED_UNIFORMS);
+
+	std::set<std::string> dupe_check;
+
+	for (U32 i = 0; i < mReservedUniforms.size(); ++i)
+	{
+		if (dupe_check.find(mReservedUniforms[i]) != dupe_check.end())
+		{
+			llerrs << "Duplicate reserved uniform name found: " << mReservedUniforms[i] << llendl;
+		}
+		dupe_check.insert(mReservedUniforms[i]);
+	}
+}
+
diff --git a/indra/llrender/llshadermgr.h b/indra/llrender/llshadermgr.h
index 2f30103811..e28bda6de2 100644
--- a/indra/llrender/llshadermgr.h
+++ b/indra/llrender/llshadermgr.h
@@ -36,9 +36,139 @@ public:
 	LLShaderMgr();
 	virtual ~LLShaderMgr();
 
+	typedef enum
+	{
+		MODELVIEW_MATRIX = 0,
+		PROJECTION_MATRIX,
+		INVERSE_PROJECTION_MATRIX,
+		MODELVIEW_PROJECTION_MATRIX,
+		NORMAL_MATRIX,
+		TEXTURE_MATRIX0,
+		TEXTURE_MATRIX1,
+		TEXTURE_MATRIX2,
+		TEXTURE_MATRIX3,
+		VIEWPORT,
+		LIGHT_POSITION,
+		LIGHT_DIRECTION,
+		LIGHT_ATTENUATION,
+		LIGHT_DIFFUSE,
+		LIGHT_AMBIENT,
+		MULTI_LIGHT_COUNT,
+		MULTI_LIGHT,
+		MULTI_LIGHT_COL,
+		MULTI_LIGHT_FAR_Z,
+		PROJECTOR_MATRIX,
+		PROJECTOR_NEAR,
+		PROJECTOR_P,
+		PROJECTOR_N,
+		PROJECTOR_ORIGIN,
+		PROJECTOR_RANGE,
+		PROJECTOR_AMBIANCE,
+		PROJECTOR_SHADOW_INDEX,
+		PROJECTOR_SHADOW_FADE,
+		PROJECTOR_FOCUS,
+		PROJECTOR_LOD,
+		PROJECTOR_AMBIENT_LOD,
+		DIFFUSE_COLOR,
+		DIFFUSE_MAP,
+		SPECULAR_MAP,
+		BUMP_MAP,
+		ENVIRONMENT_MAP,
+		CLOUD_NOISE_MAP,
+		FULLBRIGHT,
+		LIGHTNORM,
+		SUNLIGHT_COLOR,
+		AMBIENT,
+		BLUE_HORIZON,
+		BLUE_DENSITY,
+		HAZE_HORIZON,
+		HAZE_DENSITY,
+		CLOUD_SHADOW,
+		DENSITY_MULTIPLIER,
+		DISTANCE_MULTIPLIER,
+		MAX_Y,
+		GLOW,
+		CLOUD_COLOR,
+		CLOUD_POS_DENSITY1,
+		CLOUD_POS_DENSITY2,
+		CLOUD_SCALE,
+		GAMMA,
+		SCENE_LIGHT_STRENGTH,
+		LIGHT_CENTER,
+		LIGHT_SIZE,
+		LIGHT_FALLOFF,
+
+		GLOW_MIN_LUMINANCE,
+		GLOW_MAX_EXTRACT_ALPHA,
+		GLOW_LUM_WEIGHTS,
+		GLOW_WARMTH_WEIGHTS,
+		GLOW_WARMTH_AMOUNT,
+		GLOW_STRENGTH,
+		GLOW_DELTA,
+
+		MINIMUM_ALPHA,
+
+		DEFERRED_SHADOW_MATRIX,
+		DEFERRED_ENV_MAT,
+		DEFERRED_SHADOW_CLIP,
+		DEFERRED_SUN_WASH,
+		DEFERRED_SHADOW_NOISE,
+		DEFERRED_BLUR_SIZE,
+		DEFERRED_SSAO_RADIUS,
+		DEFERRED_SSAO_MAX_RADIUS,
+		DEFERRED_SSAO_FACTOR,
+		DEFERRED_SSAO_FACTOR_INV,
+		DEFERRED_SSAO_EFFECT_MAT,
+		DEFERRED_SCREEN_RES,
+		DEFERRED_NEAR_CLIP,
+		DEFERRED_SHADOW_OFFSET,
+		DEFERRED_SHADOW_BIAS,
+		DEFERRED_SPOT_SHADOW_BIAS,
+		DEFERRED_SPOT_SHADOW_OFFSET,
+		DEFERRED_SUN_DIR,
+		DEFERRED_SHADOW_RES,
+		DEFERRED_PROJ_SHADOW_RES,
+		DEFERRED_DEPTH_CUTOFF,
+		DEFERRED_NORM_CUTOFF,
+
+		FXAA_TC_SCALE,
+		FXAA_RCP_SCREEN_RES,
+		FXAA_RCP_FRAME_OPT,
+		FXAA_RCP_FRAME_OPT2,
+
+		DOF_FOCAL_DISTANCE,
+		DOF_BLUR_CONSTANT,
+		DOF_TAN_PIXEL_ANGLE,
+		DOF_MAGNIFICATION,
+		DOF_MAX_COF,
+		DOF_RES_SCALE,
+		DOF_WIDTH,
+		DOF_HEIGHT,
+
+		DEFERRED_DEPTH,
+		DEFERRED_SHADOW0,
+		DEFERRED_SHADOW1,
+		DEFERRED_SHADOW2,
+		DEFERRED_SHADOW3,
+		DEFERRED_SHADOW4,
+		DEFERRED_SHADOW5,
+		DEFERRED_NORMAL,
+		DEFERRED_POSITION,
+		DEFERRED_DIFFUSE,
+		DEFERRED_SPECULAR,
+		DEFERRED_NOISE,
+		DEFERRED_LIGHTFUNC,
+		DEFERRED_LIGHT,
+		DEFERRED_BLOOM,
+		DEFERRED_PROJECTION,
+		END_RESERVED_UNIFORMS
+	} eGLSLReservedUniforms;
+
 	// singleton pattern implementation
 	static LLShaderMgr * instance();
 
+	virtual void initAttribsAndUniforms(void);
+
 	BOOL attachShaderFeatures(LLGLSLShader * shader);
 	void dumpObjectLog(GLhandleARB ret, BOOL warns = TRUE);
 	BOOL	linkProgramObject(GLhandleARB obj, BOOL suppress_errors = FALSE);
diff --git a/indra/llrender/llvertexbuffer.cpp b/indra/llrender/llvertexbuffer.cpp
index 8fd1193780..8b5503229f 100644
--- a/indra/llrender/llvertexbuffer.cpp
+++ b/indra/llrender/llvertexbuffer.cpp
@@ -34,39 +34,50 @@
 #include "llmemtype.h"
 #include "llrender.h"
 #include "llvector4a.h"
+#include "llshadermgr.h"
 #include "llglslshader.h"
 #include "llmemory.h"
 
+//Next Highest Power Of Two
+//helper function, returns first number > v that is a power of 2, or v if v is already a power of 2
+U32 nhpo2(U32 v)
+{
+	U32 r = 1;
+	while (r < v) {
+		r *= 2;
+	}
+	return r;
+}
+
 
 //============================================================================
 
 //static
-LLVBOPool LLVertexBuffer::sStreamVBOPool;
-LLVBOPool LLVertexBuffer::sDynamicVBOPool;
-LLVBOPool LLVertexBuffer::sStreamIBOPool;
-LLVBOPool LLVertexBuffer::sDynamicIBOPool;
+LLVBOPool LLVertexBuffer::sStreamVBOPool(GL_STREAM_DRAW_ARB, GL_ARRAY_BUFFER_ARB);
+LLVBOPool LLVertexBuffer::sDynamicVBOPool(GL_DYNAMIC_DRAW_ARB, GL_ARRAY_BUFFER_ARB);
+LLVBOPool LLVertexBuffer::sStreamIBOPool(GL_STREAM_DRAW_ARB, GL_ELEMENT_ARRAY_BUFFER_ARB);
+LLVBOPool LLVertexBuffer::sDynamicIBOPool(GL_DYNAMIC_DRAW_ARB, GL_ELEMENT_ARRAY_BUFFER_ARB);
+U32 LLVBOPool::sBytesPooled = 0;
 
-LLPrivateMemoryPool* LLVertexBuffer::sPrivatePoolp = NULL ;
+LLPrivateMemoryPool* LLVertexBuffer::sPrivatePoolp = NULL;
 U32 LLVertexBuffer::sBindCount = 0;
 U32 LLVertexBuffer::sSetCount = 0;
 S32 LLVertexBuffer::sCount = 0;
 S32 LLVertexBuffer::sGLCount = 0;
 S32 LLVertexBuffer::sMappedCount = 0;
-BOOL LLVertexBuffer::sDisableVBOMapping = FALSE ;
-BOOL LLVertexBuffer::sEnableVBOs = TRUE;
+bool LLVertexBuffer::sDisableVBOMapping = false;
+bool LLVertexBuffer::sEnableVBOs = true;
 U32 LLVertexBuffer::sGLRenderBuffer = 0;
+U32 LLVertexBuffer::sGLRenderArray = 0;
 U32 LLVertexBuffer::sGLRenderIndices = 0;
 U32 LLVertexBuffer::sLastMask = 0;
-BOOL LLVertexBuffer::sVBOActive = FALSE;
-BOOL LLVertexBuffer::sIBOActive = FALSE;
+bool LLVertexBuffer::sVBOActive = false;
+bool LLVertexBuffer::sIBOActive = false;
 U32 LLVertexBuffer::sAllocatedBytes = 0;
-BOOL LLVertexBuffer::sMapped = FALSE;
-BOOL LLVertexBuffer::sUseStreamDraw = TRUE;
-BOOL LLVertexBuffer::sPreferStreamDraw = FALSE;
-S32	LLVertexBuffer::sWeight4Loc = -1;
-
-std::vector<U32> LLVertexBuffer::sDeleteList;
-
+bool LLVertexBuffer::sMapped = false;
+bool LLVertexBuffer::sUseStreamDraw = true;
+bool LLVertexBuffer::sUseVAO = false;
+bool LLVertexBuffer::sPreferStreamDraw = false;
 
 const U32 FENCE_WAIT_TIME_NANOSECONDS = 10000;  //1 ms
 
@@ -84,7 +95,7 @@ public:
 #endif
 	}
 
-	~LLGLSyncFence()
+	virtual ~LLGLSyncFence()
 	{
 #ifdef GL_ARB_sync
 		if (mSync)
@@ -122,6 +133,120 @@ public:
 
 };
 
+
+//which power of 2 is i?
+//assumes i is a power of 2 > 0
+U32 wpo2(U32 i)
+{
+	llassert(i > 0);
+	llassert(nhpo2(i) == i);
+
+	U32 r = 0;
+
+	while (i >>= 1) ++r;
+
+	return r;
+}
+
+volatile U8* LLVBOPool::allocate(U32& name, U32 size)
+{
+	llassert(nhpo2(size) == size);
+
+	U32 i = wpo2(size);
+
+	if (mFreeList.size() <= i)
+	{
+		mFreeList.resize(i+1);
+	}
+
+	volatile U8* ret = NULL;
+
+	if (mFreeList[i].empty())
+	{
+		//make a new buffer
+		glGenBuffersARB(1, &name);
+		glBindBufferARB(mType, name);
+		LLVertexBuffer::sAllocatedBytes += size;
+
+		if (LLVertexBuffer::sDisableVBOMapping || mUsage != GL_DYNAMIC_DRAW_ARB)
+		{
+			glBufferDataARB(mType, size, 0, mUsage);
+			ret = (U8*) ll_aligned_malloc_16(size);
+		}
+		else
+		{ //always use a true hint of static draw when allocating non-client-backed buffers
+			glBufferDataARB(mType, size, 0, GL_STATIC_DRAW_ARB);
+		}
+
+		glBindBufferARB(mType, 0);
+	}
+	else
+	{
+		name = mFreeList[i].front().mGLName;
+		ret = mFreeList[i].front().mClientData;
+
+		sBytesPooled -= size;
+
+		mFreeList[i].pop_front();
+	}
+
+	return ret;
+}
+
+void LLVBOPool::release(U32 name, volatile U8* buffer, U32 size)
+{
+	llassert(nhpo2(size) == size);
+
+	U32 i = wpo2(size);
+
+	llassert(mFreeList.size() > i);
+
+	Record rec;
+	rec.mGLName = name;
+	rec.mClientData = buffer;
+	
+	if (buffer == NULL)
+	{
+		glDeleteBuffersARB(1, &rec.mGLName);
+	}
+	else
+	{
+		sBytesPooled += size;
+		mFreeList[i].push_back(rec);
+	}
+}
+
+void LLVBOPool::cleanup()
+{
+	U32 size = 1;
+
+	for (U32 i = 0; i < mFreeList.size(); ++i)
+	{
+		record_list_t& l = mFreeList[i];
+
+		while (!l.empty())
+		{
+			Record& r = l.front();
+
+			glDeleteBuffersARB(1, &r.mGLName);
+
+			if (r.mClientData)
+			{
+				ll_aligned_free_16((void*) r.mClientData);
+			}
+
+			l.pop_front();
+
+			LLVertexBuffer::sAllocatedBytes -= size;
+			sBytesPooled -= size;
+		}
+
+		size *= 2;
+	}
+}
+
+
+//NOTE: each component must be AT LEAST 4 bytes in size to avoid a performance penalty on AMD hardware
 S32 LLVertexBuffer::sTypeSize[LLVertexBuffer::TYPE_MAX] =
 {
 	sizeof(LLVector4), // TYPE_VERTEX,
@@ -131,10 +256,12 @@ S32 LLVertexBuffer::sTypeSize[LLVertexBuffer::TYPE_MAX] =
 	sizeof(LLVector2), // TYPE_TEXCOORD2,
 	sizeof(LLVector2), // TYPE_TEXCOORD3,
 	sizeof(LLColor4U), // TYPE_COLOR,
+	sizeof(LLColor4U), // TYPE_EMISSIVE, only alpha is used currently
 	sizeof(LLVector4), // TYPE_BINORMAL,
 	sizeof(F32),	   // TYPE_WEIGHT,
 	sizeof(LLVector4), // TYPE_WEIGHT4,
 	sizeof(LLVector4), // TYPE_CLOTHWEIGHT,
+	sizeof(LLVector4), // TYPE_TEXTURE_INDEX (actually exists as position.w), no extra data, but stride is 16 bytes
 };
 
 U32 LLVertexBuffer::sGLMode[LLRender::NUM_MODES] = 
@@ -149,146 +276,153 @@ U32 LLVertexBuffer::sGLMode[LLRender::NUM_MODES] =
 	GL_LINE_LOOP,
 };
 
+
 //static
 void LLVertexBuffer::setupClientArrays(U32 data_mask)
 {
-	/*if (LLGLImmediate::sStarted)
-	{
-		llerrs << "Cannot use LLGLImmediate and LLVertexBuffer simultaneously!" << llendl;
-	}*/
-
 	if (sLastMask != data_mask)
 	{
-		U32 mask[] =
+		bool error = false;
+
+		if (gGLManager.mGLSLVersionMajor < 2 && gGLManager.mGLSLVersionMinor < 30)
 		{
-			MAP_VERTEX,
-			MAP_NORMAL,
-			MAP_TEXCOORD0,
-			MAP_COLOR,
-		};
-		
-		GLenum array[] =
-		{
-			GL_VERTEX_ARRAY,
-			GL_NORMAL_ARRAY,
-			GL_TEXTURE_COORD_ARRAY,
-			GL_COLOR_ARRAY,
-		};
-
-		BOOL error = FALSE;
-		for (U32 i = 0; i < 4; ++i)
-		{
-			if (sLastMask & mask[i])
-			{ //was enabled
-				if (!(data_mask & mask[i]) && i > 0)
-				{ //needs to be disabled
-					glDisableClientState(array[i]);
+			//make sure texture index is disabled
+			data_mask = data_mask & ~MAP_TEXTURE_INDEX;
+		}
+
+		if (LLGLSLShader::sNoFixedFunction)
+		{
+			for (U32 i = 0; i < TYPE_MAX; ++i)
+			{
+				S32 loc = i;
+										
+				U32 mask = 1 << i;
+
+				if (sLastMask & (1 << i))
+				{ //was enabled
+					if (!(data_mask & mask))
+					{ //needs to be disabled
+						glDisableVertexAttribArrayARB(loc);
+					}
 				}
-				else if (gDebugGL)
-				{ //needs to be enabled, make sure it was (DEBUG TEMPORARY)
-					if (i > 0 && !glIsEnabled(array[i]))
-					{
+				else 
+				{	//was disabled
+					if (data_mask & mask)
+					{ //needs to be enabled
+						glEnableVertexAttribArrayARB(loc);
+					}
+				}
+			}
+		}
+		else
+		{
+
+			GLenum array[] =
+			{
+				GL_VERTEX_ARRAY,
+				GL_NORMAL_ARRAY,
+				GL_TEXTURE_COORD_ARRAY,
+				GL_COLOR_ARRAY,
+			};
+
+			GLenum mask[] = 
+			{
+				MAP_VERTEX,
+				MAP_NORMAL,
+				MAP_TEXCOORD0,
+				MAP_COLOR
+			};
+
+
+
+			for (U32 i = 0; i < 4; ++i)
+			{
+				if (sLastMask & mask[i])
+				{ //was enabled
+					if (!(data_mask & mask[i]))
+					{ //needs to be disabled
+						glDisableClientState(array[i]);
+					}
+					else if (gDebugGL)
+					{ //needs to be enabled, make sure it was (DEBUG)
+						if (!glIsEnabled(array[i]))
+						{
+							if (gDebugSession)
+							{
+								error = true;
+								gFailLog << "Bad client state! " << array[i] << " disabled." << std::endl;
+							}
+							else
+							{
+								llerrs << "Bad client state! " << array[i] << " disabled." << llendl;
+							}
+						}
+					}
+				}
+				else 
+				{	//was disabled
+					if (data_mask & mask[i])
+					{ //needs to be enabled
+						glEnableClientState(array[i]);
+					}
+					else if (gDebugGL && glIsEnabled(array[i]))
+					{ //needs to be disabled, make sure it was (DEBUG TEMPORARY)
 						if (gDebugSession)
 						{
-							error = TRUE;
-							gFailLog << "Bad client state! " << array[i] << " disabled." << std::endl;
+							error = true;
+							gFailLog << "Bad client state! " << array[i] << " enabled." << std::endl;
 						}
 						else
 						{
-							llerrs << "Bad client state! " << array[i] << " disabled." << llendl;
+							llerrs << "Bad client state! " << array[i] << " enabled." << llendl;
 						}
 					}
 				}
 			}
-			else 
-			{	//was disabled
-				if (data_mask & mask[i] && i > 0)
-				{ //needs to be enabled
-					glEnableClientState(array[i]);
-				}
-				else if (gDebugGL && i > 0 && glIsEnabled(array[i]))
-				{ //needs to be disabled, make sure it was (DEBUG TEMPORARY)
-					if (gDebugSession)
-					{
-						error = TRUE;
-						gFailLog << "Bad client state! " << array[i] << " enabled." << std::endl;
-					}
-					else
-					{
-						llerrs << "Bad client state! " << array[i] << " enabled." << llendl;
+		
+			U32 map_tc[] = 
+			{
+				MAP_TEXCOORD1,
+				MAP_TEXCOORD2,
+				MAP_TEXCOORD3
+			};
+
+			for (U32 i = 0; i < 3; i++)
+			{
+				if (sLastMask & map_tc[i])
+				{
+					if (!(data_mask & map_tc[i]))
+					{ //disable
+						glClientActiveTextureARB(GL_TEXTURE1_ARB+i);
+						glDisableClientState(GL_TEXTURE_COORD_ARRAY);
+						glClientActiveTextureARB(GL_TEXTURE0_ARB);
 					}
 				}
+				else if (data_mask & map_tc[i])
+				{
+					glClientActiveTextureARB(GL_TEXTURE1_ARB+i);
+					glEnableClientState(GL_TEXTURE_COORD_ARRAY);
+					glClientActiveTextureARB(GL_TEXTURE0_ARB);
+				}
 			}
-		}
-
-		if (error)
-		{
-			ll_fail("LLVertexBuffer::setupClientArrays failed");
-		}
 
-		U32 map_tc[] = 
-		{
-			MAP_TEXCOORD1,
-			MAP_TEXCOORD2,
-			MAP_TEXCOORD3
-		};
-
-		for (U32 i = 0; i < 3; i++)
-		{
-			if (sLastMask & map_tc[i])
+			if (sLastMask & MAP_BINORMAL)
 			{
-				if (!(data_mask & map_tc[i]))
+				if (!(data_mask & MAP_BINORMAL))
 				{
-					glClientActiveTextureARB(GL_TEXTURE1_ARB+i);
+					glClientActiveTextureARB(GL_TEXTURE2_ARB);
 					glDisableClientState(GL_TEXTURE_COORD_ARRAY);
 					glClientActiveTextureARB(GL_TEXTURE0_ARB);
 				}
 			}
-			else if (data_mask & map_tc[i])
-			{
-				glClientActiveTextureARB(GL_TEXTURE1_ARB+i);
-				glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-				glClientActiveTextureARB(GL_TEXTURE0_ARB);
-			}
-		}
-
-		if (sLastMask & MAP_BINORMAL)
-		{
-			if (!(data_mask & MAP_BINORMAL))
+			else if (data_mask & MAP_BINORMAL)
 			{
 				glClientActiveTextureARB(GL_TEXTURE2_ARB);
-				glDisableClientState(GL_TEXTURE_COORD_ARRAY);
+				glEnableClientState(GL_TEXTURE_COORD_ARRAY);
 				glClientActiveTextureARB(GL_TEXTURE0_ARB);
 			}
 		}
-		else if (data_mask & MAP_BINORMAL)
-		{
-			glClientActiveTextureARB(GL_TEXTURE2_ARB);
-			glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-			glClientActiveTextureARB(GL_TEXTURE0_ARB);
-		}
-	
-		if (sLastMask & MAP_WEIGHT4)
-		{
-			if (sWeight4Loc < 0)
-			{
-				llerrs << "Weighting disabled but vertex buffer still bound!" << llendl;
-			}
-
-			if (!(data_mask & MAP_WEIGHT4))
-			{ //disable 4-component skin weight			
-				glDisableVertexAttribArrayARB(sWeight4Loc);
-			}
-		}
-		else if (data_mask & MAP_WEIGHT4)
-		{
-			if (sWeight4Loc >= 0)
-			{ //enable 4-component skin weight
-				glEnableVertexAttribArrayARB(sWeight4Loc);
-			}
-		}
 				
-
 		sLastMask = data_mask;
 	}
 }
@@ -296,32 +430,90 @@ void LLVertexBuffer::setupClientArrays(U32 data_mask)
 //static
 void LLVertexBuffer::drawArrays(U32 mode, const std::vector<LLVector3>& pos, const std::vector<LLVector3>& norm)
 {
+	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+	gGL.syncMatrices();
+
 	U32 count = pos.size();
 	llassert_always(norm.size() >= pos.size());
-	llassert_always(count > 0) ;
+	llassert_always(count > 0);
 
 	unbind();
 	
 	setupClientArrays(MAP_VERTEX | MAP_NORMAL);
 
-	glVertexPointer(3, GL_FLOAT, 0, pos[0].mV);
-	glNormalPointer(GL_FLOAT, 0, norm[0].mV);
+	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+
+	if (shader)
+	{
+		S32 loc = LLVertexBuffer::TYPE_VERTEX;
+		if (loc > -1)
+		{
+			glVertexAttribPointerARB(loc, 3, GL_FLOAT, GL_FALSE, 0, pos[0].mV);
+		}
+		loc = LLVertexBuffer::TYPE_NORMAL;
+		if (loc > -1)
+		{
+			glVertexAttribPointerARB(loc, 3, GL_FLOAT, GL_FALSE, 0, norm[0].mV);
+		}
+	}
+	else
+	{
+		glVertexPointer(3, GL_FLOAT, 0, pos[0].mV);
+		glNormalPointer(GL_FLOAT, 0, norm[0].mV);
+	}
 
 	glDrawArrays(sGLMode[mode], 0, count);
 }
 
+//static
+void LLVertexBuffer::drawElements(U32 mode, const LLVector4a* pos, const LLVector2* tc, S32 num_indices, const U16* indicesp)
+{
+	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+
+	gGL.syncMatrices();
+
+	U32 mask = LLVertexBuffer::MAP_VERTEX;
+	if (tc)
+	{
+		mask = mask | LLVertexBuffer::MAP_TEXCOORD0;
+	}
+
+	unbind();
+	
+	setupClientArrays(mask);
+
+	if (LLGLSLShader::sNoFixedFunction)
+	{
+		S32 loc = LLVertexBuffer::TYPE_VERTEX;
+		glVertexAttribPointerARB(loc, 3, GL_FLOAT, GL_FALSE, 16, pos);
+
+		if (tc)
+		{
+			loc = LLVertexBuffer::TYPE_TEXCOORD0;
+			glVertexAttribPointerARB(loc, 2, GL_FLOAT, GL_FALSE, 0, tc);
+		}
+	}
+	else
+	{
+		glTexCoordPointer(2, GL_FLOAT, 0, tc);
+		glVertexPointer(3, GL_FLOAT, 16, pos);
+	}
+
+	glDrawElements(sGLMode[mode], num_indices, GL_UNSIGNED_SHORT, indicesp);
+}
+
 void LLVertexBuffer::validateRange(U32 start, U32 end, U32 count, U32 indices_offset) const
 {
-	if (start >= (U32) mRequestedNumVerts ||
-	    end >= (U32) mRequestedNumVerts)
+	if (start >= (U32) mNumVerts ||
+	    end >= (U32) mNumVerts)
 	{
-		llerrs << "Bad vertex buffer draw range: [" << start << ", " << end << "] vs " << mRequestedNumVerts << llendl;
+		llerrs << "Bad vertex buffer draw range: [" << start << ", " << end << "] vs " << mNumVerts << llendl;
 	}
 
-	llassert(mRequestedNumIndices >= 0);
+	llassert(mNumIndices >= 0);
 
-	if (indices_offset >= (U32) mRequestedNumIndices ||
-	    indices_offset + count > (U32) mRequestedNumIndices)
+	if (indices_offset >= (U32) mNumIndices ||
+	    indices_offset + count > (U32) mNumIndices)
 	{
 		llerrs << "Bad index buffer draw range: [" << indices_offset << ", " << indices_offset+count << "]" << llendl;
 	}
@@ -336,23 +528,66 @@ void LLVertexBuffer::validateRange(U32 start, U32 end, U32 count, U32 indices_of
 				llerrs << "Index out of range: " << idx[i] << " not in [" << start << ", " << end << "]" << llendl;
 			}
 		}
+
+		LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+
+		if (shader && shader->mFeatures.mIndexedTextureChannels > 1)
+		{
+			LLStrider<LLVector4a> v;
+			//hack to get non-const reference
+			LLVertexBuffer* vb = (LLVertexBuffer*) this;
+			vb->getVertexStrider(v);
+
+			for (U32 i = start; i < end; i++)
+			{
+				S32 idx = (S32) (v[i][3]+0.25f);
+				if (idx < 0 || idx >= shader->mFeatures.mIndexedTextureChannels)
+				{
+					llerrs << "Bad texture index found in vertex data stream." << llendl;
+				}
+			}
+		}
 	}
 }
 
 void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const
 {
 	validateRange(start, end, count, indices_offset);
+	mMappable = false;
+	gGL.syncMatrices();
 
-	llassert(mRequestedNumVerts >= 0);
+	llassert(mNumVerts >= 0);
+	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
 
-	if (mGLIndices != sGLRenderIndices)
+	if (mGLArray)
 	{
-		llerrs << "Wrong index buffer bound." << llendl;
+		if (mGLArray != sGLRenderArray)
+		{
+			llerrs << "Wrong vertex array bound." << llendl;
+		}
+	}
+	else
+	{
+		if (mGLIndices != sGLRenderIndices)
+		{
+			llerrs << "Wrong index buffer bound." << llendl;
+		}
+
+		if (mGLBuffer != sGLRenderBuffer)
+		{
+			llerrs << "Wrong vertex buffer bound." << llendl;
+		}
 	}
 
-	if (mGLBuffer != sGLRenderBuffer)
+	if (gDebugGL && !mGLArray && useVBOs())
 	{
-		llerrs << "Wrong vertex buffer bound." << llendl;
+		GLint elem = 0;
+		glGetIntegerv(GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB, &elem);
+
+		if (elem != mGLIndices)
+		{
+			llerrs << "Wrong index buffer bound!" << llendl;
+		}
 	}
 
 	if (mode >= LLRender::NUM_MODES)
@@ -372,21 +607,35 @@ void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indi
 
 void LLVertexBuffer::draw(U32 mode, U32 count, U32 indices_offset) const
 {
-	llassert(mRequestedNumIndices >= 0);
-	if (indices_offset >= (U32) mRequestedNumIndices ||
-	    indices_offset + count > (U32) mRequestedNumIndices)
+	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+	mMappable = false;
+	gGL.syncMatrices();
+
+	llassert(mNumIndices >= 0);
+	if (indices_offset >= (U32) mNumIndices ||
+	    indices_offset + count > (U32) mNumIndices)
 	{
 		llerrs << "Bad index buffer draw range: [" << indices_offset << ", " << indices_offset+count << "]" << llendl;
 	}
 
-	if (mGLIndices != sGLRenderIndices)
+	if (mGLArray)
 	{
-		llerrs << "Wrong index buffer bound." << llendl;
+		if (mGLArray != sGLRenderArray)
+		{
+			llerrs << "Wrong vertex array bound." << llendl;
+		}
 	}
-
-	if (mGLBuffer != sGLRenderBuffer)
+	else
 	{
-		llerrs << "Wrong vertex buffer bound." << llendl;
+		if (mGLIndices != sGLRenderIndices)
+		{
+			llerrs << "Wrong index buffer bound." << llendl;
+		}
+
+		if (mGLBuffer != sGLRenderBuffer)
+		{
+			llerrs << "Wrong vertex buffer bound." << llendl;
+		}
 	}
 
 	if (mode >= LLRender::NUM_MODES)
@@ -404,16 +653,30 @@ void LLVertexBuffer::draw(U32 mode, U32 count, U32 indices_offset) const
 
 void LLVertexBuffer::drawArrays(U32 mode, U32 first, U32 count) const
 {
-	llassert(mRequestedNumVerts >= 0);
-	if (first >= (U32) mRequestedNumVerts ||
-	    first + count > (U32) mRequestedNumVerts)
+	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+	mMappable = false;
+	gGL.syncMatrices();
+	
+	llassert(mNumVerts >= 0);
+	if (first >= (U32) mNumVerts ||
+	    first + count > (U32) mNumVerts)
 	{
 		llerrs << "Bad vertex buffer draw range: [" << first << ", " << first+count << "]" << llendl;
 	}
 
-	if (mGLBuffer != sGLRenderBuffer || useVBOs() != sVBOActive)
+	if (mGLArray)
+	{
+		if (mGLArray != sGLRenderArray)
+		{
+			llerrs << "Wrong vertex array bound." << llendl;
+		}
+	}
+	else
 	{
-		llerrs << "Wrong vertex buffer bound." << llendl;
+		if (mGLBuffer != sGLRenderBuffer || useVBOs() != sVBOActive)
+		{
+			llerrs << "Wrong vertex buffer bound." << llendl;
+		}
 	}
 
 	if (mode >= LLRender::NUM_MODES)
@@ -431,38 +694,37 @@ void LLVertexBuffer::drawArrays(U32 mode, U32 first, U32 count) const
 //static
 void LLVertexBuffer::initClass(bool use_vbo, bool no_vbo_mapping)
 {
-	sEnableVBOs = use_vbo && gGLManager.mHasVertexBufferObject ;
-	if(sEnableVBOs)
-	{
-		//llassert_always(glBindBufferARB) ; //double check the extention for VBO is loaded.
+	sEnableVBOs = use_vbo && gGLManager.mHasVertexBufferObject;
+	sDisableVBOMapping = sEnableVBOs && no_vbo_mapping;
 
-		llinfos << "VBO is enabled." << llendl ;
-	}
-	else
-	{
-		llinfos << "VBO is disabled." << llendl ;
-	}
-
-	sDisableVBOMapping = sEnableVBOs && no_vbo_mapping ;
-
-	if(!sPrivatePoolp)
-	{
-		sPrivatePoolp = LLPrivateMemoryPoolManager::getInstance()->newPool(LLPrivateMemoryPool::STATIC) ;
+	if (!sPrivatePoolp)
+	{ 
+		sPrivatePoolp = LLPrivateMemoryPoolManager::getInstance()->newPool(LLPrivateMemoryPool::STATIC);
 	}
 }
 
 //static 
 void LLVertexBuffer::unbind()
 {
+	if (sGLRenderArray)
+	{
+#if GL_ARB_vertex_array_object
+		glBindVertexArray(0);
+#endif
+		sGLRenderArray = 0;
+		sGLRenderIndices = 0;
+		sIBOActive = false;
+	}
+
 	if (sVBOActive)
 	{
 		glBindBufferARB(GL_ARRAY_BUFFER_ARB, 0);
-		sVBOActive = FALSE;
+		sVBOActive = false;
 	}
 	if (sIBOActive)
 	{
 		glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, 0);
-		sIBOActive = FALSE;
+		sIBOActive = false;
 	}
 
 	sGLRenderBuffer = 0;
@@ -476,63 +738,88 @@ void LLVertexBuffer::cleanupClass()
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_CLEANUP_CLASS);
 	unbind();
-	clientCopy(); // deletes GL buffers
+	
+	sStreamIBOPool.cleanup();
+	sDynamicIBOPool.cleanup();
+	sStreamVBOPool.cleanup();
+	sDynamicVBOPool.cleanup();
 
 	if(sPrivatePoolp)
 	{
-		LLPrivateMemoryPoolManager::getInstance()->deletePool(sPrivatePoolp) ;
-		sPrivatePoolp = NULL ;
+		LLPrivateMemoryPoolManager::getInstance()->deletePool(sPrivatePoolp);
+		sPrivatePoolp = NULL;
 	}
 }
 
-void LLVertexBuffer::clientCopy(F64 max_time)
+//----------------------------------------------------------------------------
+
+S32 LLVertexBuffer::determineUsage(S32 usage)
 {
-	if (!sDeleteList.empty())
+	S32 ret_usage = usage;
+
+	if (!sEnableVBOs)
+	{
+		ret_usage = 0;
+	}
+	
+	if (ret_usage == GL_STREAM_DRAW_ARB && !sUseStreamDraw)
+	{
+		ret_usage = 0;
+	}
+	
+	if (ret_usage == GL_DYNAMIC_DRAW_ARB && sPreferStreamDraw)
 	{
-		glDeleteBuffersARB(sDeleteList.size(), (GLuint*) &(sDeleteList[0]));
-		sDeleteList.clear();
+		ret_usage = GL_STREAM_DRAW_ARB;
 	}
+	
+	if (ret_usage == 0 && LLRender::sGLCoreProfile)
+	{ //MUST use VBOs for all rendering
+		ret_usage = GL_STREAM_DRAW_ARB;
+	}
+	
+	if (ret_usage && ret_usage != GL_STREAM_DRAW_ARB)
+	{ //only stream_draw and dynamic_draw are supported when using VBOs, dynamic draw is the default
+		if (sDisableVBOMapping)
+		{ //always use stream draw if VBO mapping is disabled
+			ret_usage = GL_STREAM_DRAW_ARB;
+		}
+		else
+		{
+			ret_usage = GL_DYNAMIC_DRAW_ARB;
+		}
+	}
+	
+	return ret_usage;
 }
 
-//----------------------------------------------------------------------------
-
 LLVertexBuffer::LLVertexBuffer(U32 typemask, S32 usage) :
 	LLRefCount(),
 
 	mNumVerts(0),
 	mNumIndices(0),
-	mRequestedNumVerts(-1),
-	mRequestedNumIndices(-1),
-	mUsage(usage),
+	mAlignedOffset(0),
+	mAlignedIndexOffset(0),
+	mSize(0),
+	mIndicesSize(0),
+	mTypeMask(typemask),
+	mUsage(LLVertexBuffer::determineUsage(usage)),
 	mGLBuffer(0),
-	mGLIndices(0), 
+	mGLIndices(0),
+	mGLArray(0),
 	mMappedData(NULL),
-	mMappedIndexData(NULL), 
-	mVertexLocked(FALSE),
-	mIndexLocked(FALSE),
-	mFinal(FALSE),
-	mFilthy(FALSE),
-	mEmpty(TRUE),
-	mResized(FALSE),
-	mDynamicSize(FALSE),
+	mMappedIndexData(NULL),
+	mMappedDataUsingVBOs(false),
+	mMappedIndexDataUsingVBOs(false),
+	mVertexLocked(false),
+	mIndexLocked(false),
+	mFinal(false),
+	mEmpty(true),
+	mMappable(false),
 	mFence(NULL)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_CONSTRUCTOR);
-	mFence = NULL;
-	if (!sEnableVBOs)
-	{
-		mUsage = 0 ; 
-	}
 
-	if (mUsage == GL_STREAM_DRAW_ARB && !sUseStreamDraw)
-	{
-		mUsage = 0;
-	}
-	
-	if (mUsage == GL_DYNAMIC_DRAW_ARB && sPreferStreamDraw)
-	{
-		mUsage = GL_STREAM_DRAW_ARB;
-	}
+	mMappable = (mUsage == GL_DYNAMIC_DRAW_ARB && !sDisableVBOMapping);
 
 	//zero out offsets
 	for (U32 i = 0; i < TYPE_MAX; i++)
@@ -540,11 +827,6 @@ LLVertexBuffer::LLVertexBuffer(U32 typemask, S32 usage) :
 		mOffsets[i] = 0;
 	}
 
-	mTypeMask = typemask;
-	mSize = 0;
-	mAlignedOffset = 0;
-	mAlignedIndexOffset = 0;
-
 	sCount++;
 }
 
@@ -552,12 +834,12 @@ LLVertexBuffer::LLVertexBuffer(U32 typemask, S32 usage) :
 S32 LLVertexBuffer::calcOffsets(const U32& typemask, S32* offsets, S32 num_vertices)
 {
 	S32 offset = 0;
-	for (S32 i=0; i<TYPE_MAX; i++)
+	for (S32 i=0; i<TYPE_TEXTURE_INDEX; i++)
 	{
 		U32 mask = 1<<i;
 		if (typemask & mask)
 		{
-			if (offsets)
+			if (offsets && LLVertexBuffer::sTypeSize[i])
 			{
 				offsets[i] = offset;
 				offset += LLVertexBuffer::sTypeSize[i]*num_vertices;
@@ -566,6 +848,8 @@ S32 LLVertexBuffer::calcOffsets(const U32& typemask, S32* offsets, S32 num_verti
 		}
 	}
 
+	offsets[TYPE_TEXTURE_INDEX] = offsets[TYPE_VERTEX] + 12;
+	
 	return offset+16;
 }
 
@@ -573,7 +857,7 @@ S32 LLVertexBuffer::calcOffsets(const U32& typemask, S32* offsets, S32 num_verti
 S32 LLVertexBuffer::calcVertexSize(const U32& typemask)
 {
 	S32 size = 0;
-	for (S32 i = 0; i < TYPE_MAX; i++)
+	for (S32 i = 0; i < TYPE_TEXTURE_INDEX; i++)
 	{
 		U32 mask = 1<<i;
 		if (typemask & mask)
@@ -597,6 +881,14 @@ LLVertexBuffer::~LLVertexBuffer()
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_DESTRUCTOR);
 	destroyGLBuffer();
 	destroyGLIndices();
+
+	if (mGLArray)
+	{
+#if GL_ARB_vertex_array_object
+		glDeleteVertexArrays(1, &mGLArray);
+#endif
+	}
+
 	sCount--;
 
 	if (mFence)
@@ -606,7 +898,7 @@ LLVertexBuffer::~LLVertexBuffer()
 	
 	mFence = NULL;
 
-	llassert_always(!mMappedData && !mMappedIndexData) ;
+	llassert_always(!mMappedData && !mMappedIndexData);
 };
 
 void LLVertexBuffer::placeFence() const
@@ -635,39 +927,35 @@ void LLVertexBuffer::waitFence() const
 
 //----------------------------------------------------------------------------
 
-void LLVertexBuffer::genBuffer()
+void LLVertexBuffer::genBuffer(U32 size)
 {
+	mSize = nhpo2(size);
+
 	if (mUsage == GL_STREAM_DRAW_ARB)
 	{
-		mGLBuffer = sStreamVBOPool.allocate();
-	}
-	else if (mUsage == GL_DYNAMIC_DRAW_ARB)
-	{
-		mGLBuffer = sDynamicVBOPool.allocate();
+		mMappedData = sStreamVBOPool.allocate(mGLBuffer, mSize);
 	}
 	else
 	{
-		BOOST_STATIC_ASSERT(sizeof(mGLBuffer) == sizeof(GLuint));
-		glGenBuffersARB(1, (GLuint*)&mGLBuffer);
+		mMappedData = sDynamicVBOPool.allocate(mGLBuffer, mSize);
 	}
+	
 	sGLCount++;
 }
 
-void LLVertexBuffer::genIndices()
+void LLVertexBuffer::genIndices(U32 size)
 {
+	mIndicesSize = nhpo2(size);
+
 	if (mUsage == GL_STREAM_DRAW_ARB)
 	{
-		mGLIndices = sStreamIBOPool.allocate();
-	}
-	else if (mUsage == GL_DYNAMIC_DRAW_ARB)
-	{
-		mGLIndices = sDynamicIBOPool.allocate();
+		mMappedIndexData = sStreamIBOPool.allocate(mGLIndices, mIndicesSize);
 	}
 	else
 	{
-		BOOST_STATIC_ASSERT(sizeof(mGLBuffer) == sizeof(GLuint));
-		glGenBuffersARB(1, (GLuint*)&mGLIndices);
+		mMappedIndexData = sDynamicIBOPool.allocate(mGLIndices, mIndicesSize);
 	}
+	
 	sGLCount++;
 }
 
@@ -675,16 +963,16 @@ void LLVertexBuffer::releaseBuffer()
 {
 	if (mUsage == GL_STREAM_DRAW_ARB)
 	{
-		sStreamVBOPool.release(mGLBuffer);
-	}
-	else if (mUsage == GL_DYNAMIC_DRAW_ARB)
-	{
-		sDynamicVBOPool.release(mGLBuffer);
+		sStreamVBOPool.release(mGLBuffer, mMappedData, mSize);
 	}
 	else
 	{
-		sDeleteList.push_back(mGLBuffer);
+		sDynamicVBOPool.release(mGLBuffer, mMappedData, mSize);
 	}
+	
+	mGLBuffer = 0;
+	mMappedData = NULL;
+
 	sGLCount--;
 }
 
@@ -692,24 +980,23 @@ void LLVertexBuffer::releaseIndices()
 {
 	if (mUsage == GL_STREAM_DRAW_ARB)
 	{
-		sStreamIBOPool.release(mGLIndices);
-	}
-	else if (mUsage == GL_DYNAMIC_DRAW_ARB)
-	{
-		sDynamicIBOPool.release(mGLIndices);
+		sStreamIBOPool.release(mGLIndices, mMappedIndexData, mIndicesSize);
 	}
 	else
 	{
-		sDeleteList.push_back(mGLIndices);
+		sDynamicIBOPool.release(mGLIndices, mMappedIndexData, mIndicesSize);
 	}
+
+	mGLIndices = 0;
+	mMappedIndexData = NULL;
+	
 	sGLCount--;
 }
 
-void LLVertexBuffer::createGLBuffer()
+void LLVertexBuffer::createGLBuffer(U32 size)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_CREATE_VERTICES);
 	
-	U32 size = getSize();
 	if (mGLBuffer)
 	{
 		destroyGLBuffer();
@@ -720,27 +1007,27 @@ void LLVertexBuffer::createGLBuffer()
 		return;
 	}
 
-	mEmpty = TRUE;
+	mEmpty = true;
 
-	if (useVBOs())
+	mMappedDataUsingVBOs = useVBOs();
+	
+	if (mMappedDataUsingVBOs)
 	{
-		mMappedData = NULL;
-		genBuffer();
-		mResized = TRUE;
+		genBuffer(size);
 	}
 	else
 	{
 		static int gl_buffer_idx = 0;
 		mGLBuffer = ++gl_buffer_idx;
 		mMappedData = (U8*)ALLOCATE_MEM(sPrivatePoolp, size);
+		mSize = size;
 	}
 }
 
-void LLVertexBuffer::createGLIndices()
+void LLVertexBuffer::createGLIndices(U32 size)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_CREATE_INDICES);
-	U32 size = getIndicesSize();
-
+	
 	if (mGLIndices)
 	{
 		destroyGLIndices();
@@ -751,24 +1038,25 @@ void LLVertexBuffer::createGLIndices()
 		return;
 	}
 
-	mEmpty = TRUE;
+	mEmpty = true;
 
 	//pad by 16 bytes for aligned copies
 	size += 16;
 
-	if (useVBOs())
+	mMappedIndexDataUsingVBOs = useVBOs();
+
+	if (mMappedIndexDataUsingVBOs)
 	{
 		//pad by another 16 bytes for VBO pointer adjustment
 		size += 16;
-		mMappedIndexData = NULL;
-		genIndices();
-		mResized = TRUE;
+		genIndices(size);
 	}
 	else
 	{
 		mMappedIndexData = (U8*)ALLOCATE_MEM(sPrivatePoolp, size);
 		static int gl_buffer_idx = 0;
 		mGLIndices = ++gl_buffer_idx;
+		mIndicesSize = size;
 	}
 }
 
@@ -777,24 +1065,16 @@ void LLVertexBuffer::destroyGLBuffer()
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_DESTROY_BUFFER);
 	if (mGLBuffer)
 	{
-		if (useVBOs())
+		if (mMappedDataUsingVBOs)
 		{
-			freeClientBuffer() ;
-
-			if (mMappedData || mMappedIndexData)
-			{
-				llerrs << "Vertex buffer destroyed while mapped!" << llendl;
-			}
 			releaseBuffer();
 		}
 		else
 		{
-			FREE_MEM(sPrivatePoolp, mMappedData) ;
+			FREE_MEM(sPrivatePoolp, (void*) mMappedData);
 			mMappedData = NULL;
-			mEmpty = TRUE;
+			mEmpty = true;
 		}
-
-		sAllocatedBytes -= getSize();
 	}
 	
 	mGLBuffer = 0;
@@ -806,24 +1086,16 @@ void LLVertexBuffer::destroyGLIndices()
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_DESTROY_INDICES);
 	if (mGLIndices)
 	{
-		if (useVBOs())
+		if (mMappedIndexDataUsingVBOs)
 		{
-			freeClientBuffer() ;
-
-			if (mMappedData || mMappedIndexData)
-			{
-				llerrs << "Vertex buffer destroyed while mapped." << llendl;
-			}
 			releaseIndices();
 		}
 		else
 		{
-			FREE_MEM(sPrivatePoolp, mMappedIndexData) ;
+			FREE_MEM(sPrivatePoolp, (void*) mMappedIndexData);
 			mMappedIndexData = NULL;
-			mEmpty = TRUE;
+			mEmpty = true;
 		}
-
-		sAllocatedBytes -= getIndicesSize();
 	}
 
 	mGLIndices = 0;
@@ -842,23 +1114,14 @@ void LLVertexBuffer::updateNumVerts(S32 nverts)
 		nverts = 65535;
 	}
 
-	mRequestedNumVerts = nverts;
+	U32 needed_size = calcOffsets(mTypeMask, mOffsets, nverts);
 
-	if (!mDynamicSize)
+	if (needed_size > mSize || needed_size <= mSize/2)
 	{
-		mNumVerts = nverts;
+		createGLBuffer(needed_size);
 	}
-	else if (mUsage == GL_STATIC_DRAW_ARB ||
-		nverts > mNumVerts ||
-		nverts < mNumVerts/2)
-	{
-		if (mUsage != GL_STATIC_DRAW_ARB && nverts + nverts/4 <= 65535)
-		{
-			nverts += nverts/4;
-		}
-		mNumVerts = nverts;
-	}
-	mSize = calcOffsets(mTypeMask, mOffsets, mNumVerts);
+
+	mNumVerts = nverts;
 }
 
 void LLVertexBuffer::updateNumIndices(S32 nindices)
@@ -867,28 +1130,22 @@ void LLVertexBuffer::updateNumIndices(S32 nindices)
 
 	llassert(nindices >= 0);
 
-	mRequestedNumIndices = nindices;
-	if (!mDynamicSize)
+	U32 needed_size = sizeof(U16) * nindices;
+
+	if (needed_size > mIndicesSize || needed_size <= mIndicesSize/2)
 	{
-		mNumIndices = nindices;
+		createGLIndices(needed_size);
 	}
-	else if (mUsage == GL_STATIC_DRAW_ARB ||
-		nindices > mNumIndices ||
-		nindices < mNumIndices/2)
-	{
-		if (mUsage != GL_STATIC_DRAW_ARB)
-		{
-			nindices += nindices/4;
-		}
 
-		mNumIndices = nindices;
-	}
+	mNumIndices = nindices;
 }
 
 void LLVertexBuffer::allocateBuffer(S32 nverts, S32 nindices, bool create)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_ALLOCATE_BUFFER);
-		
+	
+	stop_glerror();
+
 	if (nverts < 0 || nindices < 0 ||
 		nverts > 65536)
 	{
@@ -898,139 +1155,169 @@ void LLVertexBuffer::allocateBuffer(S32 nverts, S32 nindices, bool create)
 	updateNumVerts(nverts);
 	updateNumIndices(nindices);
 	
-	if (mMappedData)
-	{
-		llerrs << "LLVertexBuffer::allocateBuffer() called redundantly." << llendl;
-	}
 	if (create && (nverts || nindices))
 	{
-		createGLBuffer();
-		createGLIndices();
+		//actually allocate space for the vertex buffer if using VBO mapping
+		flush();
+
+		if (gGLManager.mHasVertexArrayObject && useVBOs() && (LLRender::sGLCoreProfile || sUseVAO))
+		{
+#if GL_ARB_vertex_array_object
+			glGenVertexArrays(1, &mGLArray);
+#endif
+			setupVertexArray();
+		}
 	}
-	
-	sAllocatedBytes += getSize() + getIndicesSize();
 }
 
-void LLVertexBuffer::resizeBuffer(S32 newnverts, S32 newnindices)
-{
-	llassert(newnverts >= 0);
-	llassert(newnindices >= 0);
-
-	mRequestedNumVerts = newnverts;
-	mRequestedNumIndices = newnindices;
+static LLFastTimer::DeclareTimer FTM_SETUP_VERTEX_ARRAY("Setup VAO");
 
-	LLMemType mt2(LLMemType::MTYPE_VERTEX_RESIZE_BUFFER);
-	mDynamicSize = TRUE;
-	if (mUsage == GL_STATIC_DRAW_ARB)
-	{ //always delete/allocate static buffers on resize
-		destroyGLBuffer();
-		destroyGLIndices();
-		allocateBuffer(newnverts, newnindices, TRUE);
-		mFinal = FALSE;
-	}
-	else if (newnverts > mNumVerts || newnindices > mNumIndices ||
-			 newnverts < mNumVerts/2 || newnindices < mNumIndices/2)
+void LLVertexBuffer::setupVertexArray()
+{
+	if (!mGLArray)
 	{
-		sAllocatedBytes -= getSize() + getIndicesSize();
-		
-		updateNumVerts(newnverts);		
-		updateNumIndices(newnindices);
-		
-		S32 newsize = getSize();
-		S32 new_index_size = getIndicesSize();
-
-		sAllocatedBytes += newsize + new_index_size;
+		return;
+	}
 
-		if (newsize)
+	LLFastTimer t(FTM_SETUP_VERTEX_ARRAY);
+#if GL_ARB_vertex_array_object
+	glBindVertexArray(mGLArray);
+#endif
+	sGLRenderArray = mGLArray;
+
+	U32 attrib_size[] = 
+	{
+		3, //TYPE_VERTEX,
+		3, //TYPE_NORMAL,
+		2, //TYPE_TEXCOORD0,
+		2, //TYPE_TEXCOORD1,
+		2, //TYPE_TEXCOORD2,
+		2, //TYPE_TEXCOORD3,
+		4, //TYPE_COLOR,
+		4, //TYPE_EMISSIVE,
+		3, //TYPE_BINORMAL,
+		1, //TYPE_WEIGHT,
+		4, //TYPE_WEIGHT4,
+		4, //TYPE_CLOTHWEIGHT,
+		4, //TYPE_TEXTURE_INDEX
+	};
+
+	U32 attrib_type[] =
+	{
+		GL_FLOAT, //TYPE_VERTEX,
+		GL_FLOAT, //TYPE_NORMAL,
+		GL_FLOAT, //TYPE_TEXCOORD0,
+		GL_FLOAT, //TYPE_TEXCOORD1,
+		GL_FLOAT, //TYPE_TEXCOORD2,
+		GL_FLOAT, //TYPE_TEXCOORD3,
+		GL_UNSIGNED_BYTE, //TYPE_COLOR,
+		GL_UNSIGNED_BYTE, //TYPE_EMISSIVE,
+		GL_FLOAT,   //TYPE_BINORMAL,
+		GL_FLOAT, //TYPE_WEIGHT,
+		GL_FLOAT, //TYPE_WEIGHT4,
+		GL_FLOAT, //TYPE_CLOTHWEIGHT,
+		GL_UNSIGNED_BYTE, //TYPE_TEXTURE_INDEX
+	};
+
+	bool attrib_integer[] = 
+	{
+		false, //TYPE_VERTEX,
+		false, //TYPE_NORMAL,
+		false, //TYPE_TEXCOORD0,
+		false, //TYPE_TEXCOORD1,
+		false, //TYPE_TEXCOORD2,
+		false, //TYPE_TEXCOORD3,
+		false, //TYPE_COLOR,
+		false, //TYPE_EMISSIVE,
+		false, //TYPE_BINORMAL,
+		false, //TYPE_WEIGHT,
+		false, //TYPE_WEIGHT4,
+		false, //TYPE_CLOTHWEIGHT,
+		true, //TYPE_TEXTURE_INDEX
+	};
+
+	U32 attrib_normalized[] =
+	{
+		GL_FALSE, //TYPE_VERTEX,
+		GL_FALSE, //TYPE_NORMAL,
+		GL_FALSE, //TYPE_TEXCOORD0,
+		GL_FALSE, //TYPE_TEXCOORD1,
+		GL_FALSE, //TYPE_TEXCOORD2,
+		GL_FALSE, //TYPE_TEXCOORD3,
+		GL_TRUE, //TYPE_COLOR,
+		GL_TRUE, //TYPE_EMISSIVE,
+		GL_FALSE,   //TYPE_BINORMAL,
+		GL_FALSE, //TYPE_WEIGHT,
+		GL_FALSE, //TYPE_WEIGHT4,
+		GL_FALSE, //TYPE_CLOTHWEIGHT,
+		GL_FALSE, //TYPE_TEXTURE_INDEX
+	};
+
+	bindGLBuffer(true);
+	bindGLIndices(true);
+
+	for (U32 i = 0; i < TYPE_MAX; ++i)
+	{
+		if (mTypeMask & (1 << i))
 		{
-			if (!mGLBuffer)
-			{ //no buffer exists, create a new one
-				createGLBuffer();
-			}
-			else
+			glEnableVertexAttribArrayARB(i);
+
+			if (attrib_integer[i])
 			{
-				if (!useVBOs())
+#if !LL_DARWIN
+				//glVertexattribIPointer requires GLSL 1.30 or later
+				if (gGLManager.mGLSLVersionMajor > 1 || gGLManager.mGLSLVersionMinor >= 30)
 				{
-					FREE_MEM(sPrivatePoolp, mMappedData);
-					mMappedData = (U8*)ALLOCATE_MEM(sPrivatePoolp, newsize);
+					glVertexAttribIPointer(i, attrib_size[i], attrib_type[i], sTypeSize[i], (void*) mOffsets[i]); 
 				}
-				mResized = TRUE;
-			}
-		}
-		else if (mGLBuffer)
-		{
-			destroyGLBuffer();
-		}
-		
-		if (new_index_size)
-		{
-			if (!mGLIndices)
-			{
-				createGLIndices();
+#endif
 			}
 			else
 			{
-				if (!useVBOs())
-				{
-					FREE_MEM(sPrivatePoolp, mMappedIndexData) ;
-					mMappedIndexData = (U8*)ALLOCATE_MEM(sPrivatePoolp, new_index_size);
-				}
-				mResized = TRUE;
+				glVertexAttribPointerARB(i, attrib_size[i], attrib_type[i], attrib_normalized[i], sTypeSize[i], (void*) mOffsets[i]); 
 			}
 		}
-		else if (mGLIndices)
+		else
 		{
-			destroyGLIndices();
+			glDisableVertexAttribArrayARB(i);
 		}
 	}
 
-	if (mResized && useVBOs())
-	{
-		freeClientBuffer() ;
-		setBuffer(0);
-	}
-}
+	//draw a dummy triangle to set index array pointer
+	//glDrawElements(GL_TRIANGLES, 0, GL_UNSIGNED_SHORT, NULL);
 
-BOOL LLVertexBuffer::useVBOs() const
-{
-	//it's generally ineffective to use VBO for things that are streaming on apple
-		
-	if (!mUsage)
-	{
-		return FALSE;
-	}
-
-	return TRUE;
+	unbind();
 }
 
-//----------------------------------------------------------------------------
-void LLVertexBuffer::freeClientBuffer()
+void LLVertexBuffer::resizeBuffer(S32 newnverts, S32 newnindices)
 {
-	if(useVBOs() && sDisableVBOMapping && (mMappedData || mMappedIndexData))
-	{
-		FREE_MEM(sPrivatePoolp, mMappedData) ;
-		FREE_MEM(sPrivatePoolp, mMappedIndexData) ;
-		mMappedData = NULL ;
-		mMappedIndexData = NULL ;
-	}
-}
+	llassert(newnverts >= 0);
+	llassert(newnindices >= 0);
 
-void LLVertexBuffer::allocateClientVertexBuffer()
-{
-	if(!mMappedData)
+	LLMemType mt2(LLMemType::MTYPE_VERTEX_RESIZE_BUFFER);
+	
+	updateNumVerts(newnverts);		
+	updateNumIndices(newnindices);
+	
+	if (useVBOs())
 	{
-		mMappedData = (U8*)ALLOCATE_MEM(sPrivatePoolp, getSize());
+		flush();
+
+		if (mGLArray)
+		{ //if size changed, offsets changed
+			setupVertexArray();
+		}
 	}
 }
 
-void LLVertexBuffer::allocateClientIndexBuffer()
+bool LLVertexBuffer::useVBOs() const
 {
-	if(!mMappedIndexData)
-	{
-		mMappedIndexData = (U8*)ALLOCATE_MEM(sPrivatePoolp, getIndicesSize());		
-	}
+	//it's generally ineffective to use VBO for things that are streaming on apple
+	return (mUsage != 0);
 }
 
+//----------------------------------------------------------------------------
+
 bool expand_region(LLVertexBuffer::MappedRegion& region, S32 index, S32 count)
 {
 	S32 end = index+count;
@@ -1049,9 +1336,13 @@ bool expand_region(LLVertexBuffer::MappedRegion& region, S32 index, S32 count)
 	return true;
 }
 
+static LLFastTimer::DeclareTimer FTM_VBO_MAP_BUFFER_RANGE("VBO Map Range");
+static LLFastTimer::DeclareTimer FTM_VBO_MAP_BUFFER("VBO Map");
+
 // Map for data access
-U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range)
+volatile U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range)
 {
+	bindGLBuffer(true);
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_MAP_BUFFER);
 	if (mFinal)
 	{
@@ -1064,7 +1355,7 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 		
 	if (useVBOs())
 	{
-		if (sDisableVBOMapping || gGLManager.mHasMapBufferRange || gGLManager.mHasFlushBufferRange)
+		if (!mMappable || gGLManager.mHasMapBufferRange || gGLManager.mHasFlushBufferRange)
 		{
 			if (count == -1)
 			{
@@ -1089,7 +1380,7 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 			if (!mapped)
 			{
 				//not already mapped, map new region
-				MappedRegion region(type, !sDisableVBOMapping && map_range ? -1 : index, count);
+				MappedRegion region(type, mMappable && map_range ? -1 : index, count);
 				mMappedVertexRegions.push_back(region);
 			}
 		}
@@ -1102,25 +1393,24 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 		if (!mVertexLocked)
 		{
 			LLMemType mt_v(LLMemType::MTYPE_VERTEX_MAP_BUFFER_VERTICES);
-			setBuffer(0, type);
-			mVertexLocked = TRUE;
+			mVertexLocked = true;
 			sMappedCount++;
 			stop_glerror();	
 
-			if(sDisableVBOMapping)
+			if(!mMappable)
 			{
 				map_range = false;
-				allocateClientVertexBuffer() ;
 			}
 			else
 			{
-				U8* src = NULL;
+				volatile U8* src = NULL;
 				waitFence();
 				if (gGLManager.mHasMapBufferRange)
 				{
 					if (map_range)
 					{
 #ifdef GL_ARB_map_buffer_range
+						LLFastTimer t(FTM_VBO_MAP_BUFFER_RANGE);
 						S32 offset = mOffsets[type] + sTypeSize[type]*index;
 						S32 length = (sTypeSize[type]*count+0xF) & ~0xF;
 						src = (U8*) glMapBufferRange(GL_ARRAY_BUFFER_ARB, offset, length, 
@@ -1132,6 +1422,19 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 					else
 					{
 #ifdef GL_ARB_map_buffer_range
+
+						if (gDebugGL)
+						{
+							GLint size = 0;
+							glGetBufferParameterivARB(GL_ARRAY_BUFFER_ARB, GL_BUFFER_SIZE_ARB, &size);
+
+							if (size < mSize)
+							{
+								llerrs << "Invalid buffer size." << llendl;
+							}
+						}
+
+						LLFastTimer t(FTM_VBO_MAP_BUFFER);
 						src = (U8*) glMapBufferRange(GL_ARRAY_BUFFER_ARB, 0, mSize, 
 							GL_MAP_WRITE_BIT | 
 							GL_MAP_FLUSH_EXPLICIT_BIT);
@@ -1159,7 +1462,7 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 
 				llassert(src != NULL);
 
-				mMappedData = LL_NEXT_ALIGNED_ADDRESS<U8>(src);
+				mMappedData = LL_NEXT_ALIGNED_ADDRESS<volatile U8>(src);
 				mAlignedOffset = mMappedData - src;
 			
 				stop_glerror();
@@ -1169,17 +1472,17 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 			{
 				log_glerror();
 
-			//check the availability of memory
-			LLMemory::logMemoryInfo(TRUE) ; 
+				//check the availability of memory
+				LLMemory::logMemoryInfo(true);
 			
-				if(!sDisableVBOMapping)
+				if(mMappable)
 				{			
 					//--------------------
 					//print out more debug info before crash
-					llinfos << "vertex buffer size: (num verts : num indices) = " << getNumVerts() << " : " << getNumIndices() << llendl ;
-					GLint size ;
-					glGetBufferParameterivARB(GL_ARRAY_BUFFER_ARB, GL_BUFFER_SIZE_ARB, &size) ;
-					llinfos << "GL_ARRAY_BUFFER_ARB size is " << size << llendl ;
+					llinfos << "vertex buffer size: (num verts : num indices) = " << getNumVerts() << " : " << getNumIndices() << llendl;
+					GLint size;
+					glGetBufferParameterivARB(GL_ARRAY_BUFFER_ARB, GL_BUFFER_SIZE_ARB, &size);
+					llinfos << "GL_ARRAY_BUFFER_ARB size is " << size << llendl;
 					//--------------------
 
 					GLint buff;
@@ -1194,7 +1497,7 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 				}
 				else
 				{
-					llerrs << "memory allocation for vertex data failed." << llendl ;
+					llerrs << "memory allocation for vertex data failed." << llendl;
 				}
 			}
 		}
@@ -1204,7 +1507,7 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 		map_range = false;
 	}
 	
-	if (map_range && gGLManager.mHasMapBufferRange && !sDisableVBOMapping)
+	if (map_range && gGLManager.mHasMapBufferRange && mMappable)
 	{
 		return mMappedData;
 	}
@@ -1214,9 +1517,14 @@ U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_ran
 	}
 }
 
-U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
+
+static LLFastTimer::DeclareTimer FTM_VBO_MAP_INDEX_RANGE("IBO Map Range");
+static LLFastTimer::DeclareTimer FTM_VBO_MAP_INDEX("IBO Map");
+
+volatile U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_MAP_BUFFER);
+	bindGLIndices(true);
 	if (mFinal)
 	{
 		llerrs << "LLVertexBuffer::mapIndexBuffer() called on a finalized buffer." << llendl;
@@ -1228,7 +1536,7 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 
 	if (useVBOs())
 	{
-		if (sDisableVBOMapping || gGLManager.mHasMapBufferRange || gGLManager.mHasFlushBufferRange)
+		if (!mMappable || gGLManager.mHasMapBufferRange || gGLManager.mHasFlushBufferRange)
 		{
 			if (count == -1)
 			{
@@ -1250,7 +1558,7 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 			if (!mapped)
 			{
 				//not already mapped, map new region
-				MappedRegion region(TYPE_INDEX, !sDisableVBOMapping && map_range ? -1 : index, count);
+				MappedRegion region(TYPE_INDEX, mMappable && map_range ? -1 : index, count);
 				mMappedIndexRegions.push_back(region);
 			}
 		}
@@ -1264,25 +1572,35 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 		{
 			LLMemType mt_v(LLMemType::MTYPE_VERTEX_MAP_BUFFER_INDICES);
 
-			setBuffer(0, TYPE_INDEX);
-			mIndexLocked = TRUE;
+			mIndexLocked = true;
 			sMappedCount++;
 			stop_glerror();	
 
-			if(sDisableVBOMapping)
+			if (gDebugGL && useVBOs())
+			{
+				GLint elem = 0;
+				glGetIntegerv(GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB, &elem);
+
+				if (elem != mGLIndices)
+				{
+					llerrs << "Wrong index buffer bound!" << llendl;
+				}
+			}
+
+			if(!mMappable)
 			{
 				map_range = false;
-				allocateClientIndexBuffer() ;
 			}
 			else
 			{
-				U8* src = NULL;
+				volatile U8* src = NULL;
 				waitFence();
 				if (gGLManager.mHasMapBufferRange)
 				{
 					if (map_range)
 					{
 #ifdef GL_ARB_map_buffer_range
+						LLFastTimer t(FTM_VBO_MAP_INDEX_RANGE);
 						S32 offset = sizeof(U16)*index;
 						S32 length = sizeof(U16)*count;
 						src = (U8*) glMapBufferRange(GL_ELEMENT_ARRAY_BUFFER_ARB, offset, length, 
@@ -1294,6 +1612,7 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 					else
 					{
 #ifdef GL_ARB_map_buffer_range
+						LLFastTimer t(FTM_VBO_MAP_INDEX);
 						src = (U8*) glMapBufferRange(GL_ELEMENT_ARRAY_BUFFER_ARB, 0, sizeof(U16)*mNumIndices, 
 							GL_MAP_WRITE_BIT | 
 							GL_MAP_FLUSH_EXPLICIT_BIT);
@@ -1315,6 +1634,7 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 				}
 				else
 				{
+					LLFastTimer t(FTM_VBO_MAP_INDEX);
 					map_range = false;
 					src = (U8*) glMapBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
 				}
@@ -1331,9 +1651,9 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 		if (!mMappedIndexData)
 		{
 			log_glerror();
-			LLMemory::logMemoryInfo(TRUE) ;
+			LLMemory::logMemoryInfo(true);
 
-			if(!sDisableVBOMapping)
+			if(mMappable)
 			{
 				GLint buff;
 				glGetIntegerv(GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB, &buff);
@@ -1346,7 +1666,7 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 			}
 			else
 			{
-				llerrs << "memory allocation for Index data failed. " << llendl ;
+				llerrs << "memory allocation for Index data failed. " << llendl;
 			}
 		}
 	}
@@ -1355,7 +1675,7 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 		map_range = false;
 	}
 
-	if (map_range && gGLManager.mHasMapBufferRange && !sDisableVBOMapping)
+	if (map_range && gGLManager.mHasMapBufferRange && mMappable)
 	{
 		return mMappedIndexData;
 	}
@@ -1365,21 +1685,30 @@ U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 	}
 }
 
-void LLVertexBuffer::unmapBuffer(S32 type)
+static LLFastTimer::DeclareTimer FTM_VBO_UNMAP("VBO Unmap");
+static LLFastTimer::DeclareTimer FTM_VBO_FLUSH_RANGE("Flush VBO Range");
+
+
+static LLFastTimer::DeclareTimer FTM_IBO_UNMAP("IBO Unmap");
+static LLFastTimer::DeclareTimer FTM_IBO_FLUSH_RANGE("Flush IBO Range");
+
+void LLVertexBuffer::unmapBuffer()
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_UNMAP_BUFFER);
-	if (!useVBOs() || type == -2)
+	if (!useVBOs())
 	{
-		return ; //nothing to unmap
+		return; //nothing to unmap
 	}
 
-	bool updated_all = false ;
+	bool updated_all = false;
 
-	if (mMappedData && mVertexLocked && type != TYPE_INDEX)
+	if (mMappedData && mVertexLocked)
 	{
-		updated_all = (mIndexLocked && type < 0) ; //both vertex and index buffers done updating
+		LLFastTimer t(FTM_VBO_UNMAP);
+		bindGLBuffer(true);
+		updated_all = mIndexLocked; //both vertex and index buffers done updating
 
-		if(sDisableVBOMapping)
+		if(!mMappable)
 		{
 			if (!mMappedVertexRegions.empty())
 			{
@@ -1389,7 +1718,7 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 					const MappedRegion& region = mMappedVertexRegions[i];
 					S32 offset = region.mIndex >= 0 ? mOffsets[region.mType]+sTypeSize[region.mType]*region.mIndex : 0;
 					S32 length = sTypeSize[region.mType]*region.mCount;
-					glBufferSubDataARB(GL_ARRAY_BUFFER_ARB, offset, length, mMappedData+offset);
+					glBufferSubDataARB(GL_ARRAY_BUFFER_ARB, offset, length, (U8*) mMappedData+offset);
 					stop_glerror();
 				}
 
@@ -1398,7 +1727,7 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 			else
 			{
 				stop_glerror();
-				glBufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, getSize(), mMappedData);
+				glBufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, getSize(), (U8*) mMappedData);
 				stop_glerror();
 			}
 		}
@@ -1416,6 +1745,7 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 						S32 length = sTypeSize[region.mType]*region.mCount;
 						if (gGLManager.mHasMapBufferRange)
 						{
+							LLFastTimer t(FTM_VBO_FLUSH_RANGE);
 #ifdef GL_ARB_map_buffer_range
 							glFlushMappedBufferRange(GL_ARRAY_BUFFER_ARB, offset, length);
 #endif
@@ -1437,13 +1767,15 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 			mMappedData = NULL;
 		}
 
-		mVertexLocked = FALSE ;
+		mVertexLocked = false;
 		sMappedCount--;
 	}
 	
-	if (mMappedIndexData && mIndexLocked && (type < 0 || type == TYPE_INDEX))
+	if (mMappedIndexData && mIndexLocked)
 	{
-		if(sDisableVBOMapping)
+		LLFastTimer t(FTM_IBO_UNMAP);
+		bindGLIndices();
+		if(!mMappable)
 		{
 			if (!mMappedIndexRegions.empty())
 			{
@@ -1452,7 +1784,7 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 					const MappedRegion& region = mMappedIndexRegions[i];
 					S32 offset = region.mIndex >= 0 ? sizeof(U16)*region.mIndex : 0;
 					S32 length = sizeof(U16)*region.mCount;
-					glBufferSubDataARB(GL_ELEMENT_ARRAY_BUFFER_ARB, offset, length, mMappedIndexData+offset);
+					glBufferSubDataARB(GL_ELEMENT_ARRAY_BUFFER_ARB, offset, length, (U8*) mMappedIndexData+offset);
 					stop_glerror();
 				}
 
@@ -1461,7 +1793,7 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 			else
 			{
 				stop_glerror();
-				glBufferSubDataARB(GL_ELEMENT_ARRAY_BUFFER_ARB, 0, getIndicesSize(), mMappedIndexData);
+				glBufferSubDataARB(GL_ELEMENT_ARRAY_BUFFER_ARB, 0, getIndicesSize(), (U8*) mMappedIndexData);
 				stop_glerror();
 			}
 		}
@@ -1478,6 +1810,7 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 						S32 length = sizeof(U16)*region.mCount;
 						if (gGLManager.mHasMapBufferRange)
 						{
+							LLFastTimer t(FTM_IBO_FLUSH_RANGE);
 #ifdef GL_ARB_map_buffer_range
 							glFlushMappedBufferRange(GL_ELEMENT_ARRAY_BUFFER_ARB, offset, length);
 #endif
@@ -1498,30 +1831,16 @@ void LLVertexBuffer::unmapBuffer(S32 type)
 			glUnmapBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB);
 			stop_glerror();
 
-			mMappedIndexData = NULL ;
+			mMappedIndexData = NULL;
 		}
 
-		mIndexLocked = FALSE ;
+		mIndexLocked = false;
 		sMappedCount--;
 	}
 
 	if(updated_all)
 	{
-		if(mUsage == GL_STATIC_DRAW_ARB)
-		{
-			//static draw buffers can only be mapped a single time
-			//throw out client data (we won't be using it again)
-			mEmpty = TRUE;
-			mFinal = TRUE;
-			if(sDisableVBOMapping)
-			{
-				freeClientBuffer() ;
-			}
-		}
-		else
-		{
-			mEmpty = FALSE;
-		}
+		mEmpty = false;
 	}
 }
 
@@ -1536,39 +1855,39 @@ template <class T,S32 type> struct VertexBufferStrider
 	{
 		if (type == LLVertexBuffer::TYPE_INDEX)
 		{
-			U8* ptr = vbo.mapIndexBuffer(index, count, map_range);
+			volatile U8* ptr = vbo.mapIndexBuffer(index, count, map_range);
 
 			if (ptr == NULL)
 			{
 				llwarns << "mapIndexBuffer failed!" << llendl;
-				return FALSE;
+				return false;
 			}
 
 			strider = (T*)ptr;
 			strider.setStride(0);
-			return TRUE;
+			return true;
 		}
 		else if (vbo.hasDataType(type))
 		{
 			S32 stride = LLVertexBuffer::sTypeSize[type];
 
-			U8* ptr = vbo.mapVertexBuffer(type, index, count, map_range);
+			volatile U8* ptr = vbo.mapVertexBuffer(type, index, count, map_range);
 
 			if (ptr == NULL)
 			{
 				llwarns << "mapVertexBuffer failed!" << llendl;
-				return FALSE;
+				return false;
 			}
 
 			strider = (T*)ptr;
 			strider.setStride(stride);
-			return TRUE;
+			return true;
 		}
 		else
 		{
 			llerrs << "VertexBufferStrider could not find valid vertex data." << llendl;
 		}
-		return FALSE;
+		return false;
 	}
 };
 
@@ -1576,6 +1895,10 @@ bool LLVertexBuffer::getVertexStrider(LLStrider<LLVector3>& strider, S32 index,
 {
 	return VertexBufferStrider<LLVector3,TYPE_VERTEX>::get(*this, strider, index, count, map_range);
 }
+bool LLVertexBuffer::getVertexStrider(LLStrider<LLVector4a>& strider, S32 index, S32 count, bool map_range)
+{
+	return VertexBufferStrider<LLVector4a,TYPE_VERTEX>::get(*this, strider, index, count, map_range);
+}
 bool LLVertexBuffer::getIndexStrider(LLStrider<U16>& strider, S32 index, S32 count, bool map_range)
 {
 	return VertexBufferStrider<U16,TYPE_INDEX>::get(*this, strider, index, count, map_range);
@@ -1601,6 +1924,10 @@ bool LLVertexBuffer::getColorStrider(LLStrider<LLColor4U>& strider, S32 index, S
 {
 	return VertexBufferStrider<LLColor4U,TYPE_COLOR>::get(*this, strider, index, count, map_range);
 }
+bool LLVertexBuffer::getEmissiveStrider(LLStrider<LLColor4U>& strider, S32 index, S32 count, bool map_range)
+{
+	return VertexBufferStrider<LLColor4U,TYPE_EMISSIVE>::get(*this, strider, index, count, map_range);
+}
 bool LLVertexBuffer::getWeightStrider(LLStrider<F32>& strider, S32 index, S32 count, bool map_range)
 {
 	return VertexBufferStrider<F32,TYPE_WEIGHT>::get(*this, strider, index, count, map_range);
@@ -1618,43 +1945,147 @@ bool LLVertexBuffer::getClothWeightStrider(LLStrider<LLVector4>& strider, S32 in
 
 //----------------------------------------------------------------------------
 
+static LLFastTimer::DeclareTimer FTM_BIND_GL_ARRAY("Bind Array");
+bool LLVertexBuffer::bindGLArray()
+{
+	if (mGLArray && sGLRenderArray != mGLArray)
+	{
+		{
+			LLFastTimer t(FTM_BIND_GL_ARRAY);
+#if GL_ARB_vertex_array_object
+			glBindVertexArray(mGLArray);
+#endif
+			sGLRenderArray = mGLArray;
+		}
+
+		//really shouldn't be necessary, but some drivers don't properly restore the
+		//state of GL_ELEMENT_ARRAY_BUFFER_BINDING
+		bindGLIndices();
+		
+		return true;
+	}
+		
+	return false;
+}
+
+static LLFastTimer::DeclareTimer FTM_BIND_GL_BUFFER("Bind Buffer");
+
+bool LLVertexBuffer::bindGLBuffer(bool force_bind)
+{
+	bindGLArray();
+
+	bool ret = false;
+
+	if (useVBOs() && (force_bind || (mGLBuffer && (mGLBuffer != sGLRenderBuffer || !sVBOActive))))
+	{
+		LLFastTimer t(FTM_BIND_GL_BUFFER);
+		/*if (sMapped)
+		{
+			llerrs << "VBO bound while another VBO mapped!" << llendl;
+		}*/
+		glBindBufferARB(GL_ARRAY_BUFFER_ARB, mGLBuffer);
+		sGLRenderBuffer = mGLBuffer;
+		sBindCount++;
+		sVBOActive = true;
+
+		if (mGLArray)
+		{
+			llassert(sGLRenderArray == mGLArray);
+			//mCachedRenderBuffer = mGLBuffer;
+		}
+
+		ret = true;
+	}
+
+	return ret;
+}
+
+static LLFastTimer::DeclareTimer FTM_BIND_GL_INDICES("Bind Indices");
+
+bool LLVertexBuffer::bindGLIndices(bool force_bind)
+{
+	bindGLArray();
+
+	bool ret = false;
+	if (useVBOs() && (force_bind || (mGLIndices && (mGLIndices != sGLRenderIndices || !sIBOActive))))
+	{
+		LLFastTimer t(FTM_BIND_GL_INDICES);
+		/*if (sMapped)
+		{
+			llerrs << "VBO bound while another VBO mapped!" << llendl;
+		}*/
+		glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, mGLIndices);
+		sGLRenderIndices = mGLIndices;
+		stop_glerror();
+		sBindCount++;
+		sIBOActive = true;
+		ret = true;
+	}
+
+	return ret;
+}
+
+void LLVertexBuffer::flush()
+{
+	if (useVBOs())
+	{
+		unmapBuffer();
+	}
+}
+
 // Set for rendering
-void LLVertexBuffer::setBuffer(U32 data_mask, S32 type)
+void LLVertexBuffer::setBuffer(U32 data_mask)
 {
+	flush();
+
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_SET_BUFFER);
 	//set up pointers if the data mask is different ...
-	BOOL setup = (sLastMask != data_mask);
+	bool setup = (sLastMask != data_mask);
+
+	if (gDebugGL && data_mask != 0)
+	{ //make sure data requirements are fulfilled
+		LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
+		if (shader)
+		{
+			U32 required_mask = 0;
+			for (U32 i = 0; i < LLVertexBuffer::TYPE_TEXTURE_INDEX; ++i)
+			{
+				if (shader->getAttribLocation(i) > -1)
+				{
+					U32 required = 1 << i;
+					if ((data_mask & required) == 0)
+					{
+						llwarns << "Missing attribute: " << LLShaderMgr::instance()->mReservedAttribs[i] << llendl;
+					}
+
+					required_mask |= required;
+				}
+			}
+
+			if ((data_mask & required_mask) != required_mask)
+			{
+				llerrs << "Shader consumption mismatches data provision." << llendl;
+			}
+		}
+	}
 
 	if (useVBOs())
 	{
-		if (mGLBuffer && (mGLBuffer != sGLRenderBuffer || !sVBOActive))
+		if (mGLArray)
 		{
-			/*if (sMapped)
-			{
-				llerrs << "VBO bound while another VBO mapped!" << llendl;
-			}*/
-			stop_glerror();
-			glBindBufferARB(GL_ARRAY_BUFFER_ARB, mGLBuffer);
-			stop_glerror();
-			sBindCount++;
-			sVBOActive = TRUE;
-			setup = TRUE; // ... or the bound buffer changed
+			bindGLArray();
+			setup = false; //do NOT perform pointer setup if using VAO
 		}
-		if (mGLIndices && (mGLIndices != sGLRenderIndices || !sIBOActive))
+		else
 		{
-			/*if (sMapped)
-			{
-				llerrs << "VBO bound while another VBO mapped!" << llendl;
-			}*/
-			stop_glerror();
-			glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, mGLIndices);
-			stop_glerror();
-			sBindCount++;
-			sIBOActive = TRUE;
+			const bool bindBuffer = bindGLBuffer();
+			const bool bindIndices = bindGLIndices();
+			
+			setup = setup || bindBuffer || bindIndices;
 		}
-		
-		BOOL error = FALSE;
-		if (gDebugGL)
+
+		bool error = false;
+		if (gDebugGL && !mGLArray)
 		{
 			GLint buff;
 			glGetIntegerv(GL_ARRAY_BUFFER_BINDING_ARB, &buff);
@@ -1662,7 +2093,7 @@ void LLVertexBuffer::setBuffer(U32 data_mask, S32 type)
 			{
 				if (gDebugSession)
 				{
-					error = TRUE;
+					error = true;
 					gFailLog << "Invalid GL vertex buffer bound: " << buff << std::endl;
 				}
 				else
@@ -1678,7 +2109,7 @@ void LLVertexBuffer::setBuffer(U32 data_mask, S32 type)
 				{
 					if (gDebugSession)
 					{
-						error = TRUE;
+						error = true;
 						gFailLog << "Invalid GL index buffer bound: " << buff <<  std::endl;
 					}
 					else
@@ -1689,116 +2120,55 @@ void LLVertexBuffer::setBuffer(U32 data_mask, S32 type)
 			}
 		}
 
-		if (mResized)
+		
+	}
+	else
+	{	
+		if (sGLRenderArray)
 		{
-			if (gDebugGL)
-			{
-				GLint buff;
-				glGetIntegerv(GL_ARRAY_BUFFER_BINDING_ARB, &buff);
-				if ((GLuint)buff != mGLBuffer)
-				{
-					if (gDebugSession)
-					{
-						error = TRUE;
-						gFailLog << "Invalid GL vertex buffer bound: " << std::endl;
-					}
-					else
-					{
-						llerrs << "Invalid GL vertex buffer bound: " << buff << llendl;
-					}
-				}
-
-				if (mGLIndices != 0)
-				{
-					glGetIntegerv(GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB, &buff);
-					if ((GLuint)buff != mGLIndices)
-					{
-						if (gDebugSession)
-						{
-							error = TRUE;
-							gFailLog << "Invalid GL index buffer bound: "<< std::endl;
-						}
-						else
-						{
-							llerrs << "Invalid GL index buffer bound: " << buff << llendl;
-						}
-					}
-				}
-			}
-
-			if (mGLBuffer)
-			{
-				stop_glerror();
-				glBufferDataARB(GL_ARRAY_BUFFER_ARB, getSize(), NULL, mUsage);
-				stop_glerror();
-			}
-			if (mGLIndices)
-			{
-				stop_glerror();
-				glBufferDataARB(GL_ELEMENT_ARRAY_BUFFER_ARB, getIndicesSize(), NULL, mUsage);
-				stop_glerror();
-			}
-
-			mEmpty = TRUE;
-			mResized = FALSE;
-
-			if (data_mask != 0)
-			{
-				if (gDebugSession)
-				{
-					error = TRUE;
-					gFailLog << "Buffer set for rendering before being filled after resize." << std::endl;
-				}
-				else
-				{
-					llerrs << "Buffer set for rendering before being filled after resize." << llendl;
-				}
-			}
+#if GL_ARB_vertex_array_object
+			glBindVertexArray(0);
+#endif
+			sGLRenderArray = 0;
+			sGLRenderIndices = 0;
+			sIBOActive = false;
 		}
 
-		if (error)
-		{
-			ll_fail("LLVertexBuffer::mapBuffer failed");
-		}
-		unmapBuffer(type);
-	}
-	else
-	{		
 		if (mGLBuffer)
 		{
 			if (sVBOActive)
 			{
 				glBindBufferARB(GL_ARRAY_BUFFER_ARB, 0);
 				sBindCount++;
-				sVBOActive = FALSE;
-				setup = TRUE; // ... or a VBO is deactivated
+				sVBOActive = false;
+				setup = true; // ... or a VBO is deactivated
 			}
 			if (sGLRenderBuffer != mGLBuffer)
 			{
-				setup = TRUE; // ... or a client memory pointer changed
+				sGLRenderBuffer = mGLBuffer;
+				setup = true; // ... or a client memory pointer changed
 			}
 		}
-		if (mGLIndices && sIBOActive)
+		if (mGLIndices)
 		{
-			/*if (sMapped)
+			if (sIBOActive)
 			{
-				llerrs << "VBO unbound while potentially mapped!" << llendl;
-			}*/
-			glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, 0);
-			sBindCount++;
-			sIBOActive = FALSE;
+				glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, 0);
+				sBindCount++;
+				sIBOActive = false;
+			}
+			
+			sGLRenderIndices = mGLIndices;
 		}
 	}
 
-	setupClientArrays(data_mask);
-	
-	if (mGLIndices)
+	if (!mGLArray)
 	{
-		sGLRenderIndices = mGLIndices;
+		setupClientArrays(data_mask);
 	}
+			
 	if (mGLBuffer)
 	{
-		sGLRenderBuffer = mGLBuffer;
 		if (data_mask && setup)
 		{
 			setupVertexBuffer(data_mask); // subclass specific setup (virtual function)
@@ -1808,80 +2178,153 @@ void LLVertexBuffer::setBuffer(U32 data_mask, S32 type)
 }
 
 // virtual (default)
-void LLVertexBuffer::setupVertexBuffer(U32 data_mask) const
+void LLVertexBuffer::setupVertexBuffer(U32 data_mask)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_SETUP_VERTEX_BUFFER);
 	stop_glerror();
-	U8* base = useVBOs() ? (U8*) mAlignedOffset : mMappedData;
+	volatile U8* base = useVBOs() ? (U8*) mAlignedOffset : mMappedData;
 
-	if ((data_mask & mTypeMask) != data_mask)
+	/*if ((data_mask & mTypeMask) != data_mask)
 	{
 		llerrs << "LLVertexBuffer::setupVertexBuffer missing required components for supplied data mask." << llendl;
-	}
-
-	if (data_mask & MAP_NORMAL)
-	{
-		glNormalPointer(GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_NORMAL], (void*)(base + mOffsets[TYPE_NORMAL]));
-	}
-	if (data_mask & MAP_TEXCOORD3)
-	{
-		glClientActiveTextureARB(GL_TEXTURE3_ARB);
-		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD3], (void*)(base + mOffsets[TYPE_TEXCOORD3]));
-		glClientActiveTextureARB(GL_TEXTURE0_ARB);
-	}
-	if (data_mask & MAP_TEXCOORD2)
-	{
-		glClientActiveTextureARB(GL_TEXTURE2_ARB);
-		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD2], (void*)(base + mOffsets[TYPE_TEXCOORD2]));
-		glClientActiveTextureARB(GL_TEXTURE0_ARB);
-	}
-	if (data_mask & MAP_TEXCOORD1)
-	{
-		glClientActiveTextureARB(GL_TEXTURE1_ARB);
-		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD1], (void*)(base + mOffsets[TYPE_TEXCOORD1]));
-		glClientActiveTextureARB(GL_TEXTURE0_ARB);
-	}
-	if (data_mask & MAP_BINORMAL)
-	{
-		glClientActiveTextureARB(GL_TEXTURE2_ARB);
-		glTexCoordPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_BINORMAL], (void*)(base + mOffsets[TYPE_BINORMAL]));
-		glClientActiveTextureARB(GL_TEXTURE0_ARB);
-	}
-	if (data_mask & MAP_TEXCOORD0)
-	{
-		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD0], (void*)(base + mOffsets[TYPE_TEXCOORD0]));
-	}
-	if (data_mask & MAP_COLOR)
-	{
-		glColorPointer(4, GL_UNSIGNED_BYTE, LLVertexBuffer::sTypeSize[TYPE_COLOR], (void*)(base + mOffsets[TYPE_COLOR]));
-	}
-	
-	if (data_mask & MAP_WEIGHT)
-	{
-		glVertexAttribPointerARB(1, 1, GL_FLOAT, FALSE, LLVertexBuffer::sTypeSize[TYPE_WEIGHT], (void*)(base + mOffsets[TYPE_WEIGHT]));
-	}
-
-	if (data_mask & MAP_WEIGHT4 && sWeight4Loc != -1)
-	{
-		glVertexAttribPointerARB(sWeight4Loc, 4, GL_FLOAT, FALSE, LLVertexBuffer::sTypeSize[TYPE_WEIGHT4], (void*)(base+mOffsets[TYPE_WEIGHT4]));
-	}
+	}*/
 
-	if (data_mask & MAP_CLOTHWEIGHT)
+	if (LLGLSLShader::sNoFixedFunction)
 	{
-		glVertexAttribPointerARB(4, 4, GL_FLOAT, TRUE,  LLVertexBuffer::sTypeSize[TYPE_CLOTHWEIGHT], (void*)(base + mOffsets[TYPE_CLOTHWEIGHT]));
-	}
-	if (data_mask & MAP_VERTEX)
+		if (data_mask & MAP_NORMAL)
+		{
+			S32 loc = TYPE_NORMAL;
+			void* ptr = (void*)(base + mOffsets[TYPE_NORMAL]);
+			glVertexAttribPointerARB(loc, 3, GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_NORMAL], ptr);
+		}
+		if (data_mask & MAP_TEXCOORD3)
+		{
+			S32 loc = TYPE_TEXCOORD3;
+			void* ptr = (void*)(base + mOffsets[TYPE_TEXCOORD3]);
+			glVertexAttribPointerARB(loc,2,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD3], ptr);
+		}
+		if (data_mask & MAP_TEXCOORD2)
+		{
+			S32 loc = TYPE_TEXCOORD2;
+			void* ptr = (void*)(base + mOffsets[TYPE_TEXCOORD2]);
+			glVertexAttribPointerARB(loc,2,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD2], ptr);
+		}
+		if (data_mask & MAP_TEXCOORD1)
+		{
+			S32 loc = TYPE_TEXCOORD1;
+			void* ptr = (void*)(base + mOffsets[TYPE_TEXCOORD1]);
+			glVertexAttribPointerARB(loc,2,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD1], ptr);
+		}
+		if (data_mask & MAP_BINORMAL)
+		{
+			S32 loc = TYPE_BINORMAL;
+			void* ptr = (void*)(base + mOffsets[TYPE_BINORMAL]);
+			glVertexAttribPointerARB(loc, 3,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_BINORMAL], ptr);
+		}
+		if (data_mask & MAP_TEXCOORD0)
+		{
+			S32 loc = TYPE_TEXCOORD0;
+			void* ptr = (void*)(base + mOffsets[TYPE_TEXCOORD0]);
+			glVertexAttribPointerARB(loc,2,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD0], ptr);
+		}
+		if (data_mask & MAP_COLOR)
+		{
+			S32 loc = TYPE_COLOR;
+			void* ptr = (void*)(base + mOffsets[TYPE_COLOR]);
+			glVertexAttribPointerARB(loc, 4, GL_UNSIGNED_BYTE, GL_TRUE, LLVertexBuffer::sTypeSize[TYPE_COLOR], ptr);
+		}
+		if (data_mask & MAP_EMISSIVE)
+		{
+			S32 loc = TYPE_EMISSIVE;
+			void* ptr = (void*)(base + mOffsets[TYPE_EMISSIVE]);
+			glVertexAttribPointerARB(loc, 4, GL_UNSIGNED_BYTE, GL_TRUE, LLVertexBuffer::sTypeSize[TYPE_EMISSIVE], ptr);
+		}
+		if (data_mask & MAP_WEIGHT)
+		{
+			S32 loc = TYPE_WEIGHT;
+			void* ptr = (void*)(base + mOffsets[TYPE_WEIGHT]);
+			glVertexAttribPointerARB(loc, 1, GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_WEIGHT], ptr);
+		}
+		if (data_mask & MAP_WEIGHT4)
+		{
+			S32 loc = TYPE_WEIGHT4;
+			void* ptr = (void*)(base+mOffsets[TYPE_WEIGHT4]);
+			glVertexAttribPointerARB(loc, 4, GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_WEIGHT4], ptr);
+		}
+		if (data_mask & MAP_CLOTHWEIGHT)
+		{
+			S32 loc = TYPE_CLOTHWEIGHT;
+			void* ptr = (void*)(base + mOffsets[TYPE_CLOTHWEIGHT]);
+			glVertexAttribPointerARB(loc, 4, GL_FLOAT, GL_TRUE,  LLVertexBuffer::sTypeSize[TYPE_CLOTHWEIGHT], ptr);
+		}
+		if (data_mask & MAP_TEXTURE_INDEX && 
+				(gGLManager.mGLSLVersionMajor >= 2 || gGLManager.mGLSLVersionMinor >= 30)) //indexed texture rendering requires GLSL 1.30 or later
+		{
+#if !LL_DARWIN
+			S32 loc = TYPE_TEXTURE_INDEX;
+			void *ptr = (void*) (base + mOffsets[TYPE_VERTEX] + 12);
+			glVertexAttribIPointer(loc, 4, GL_UNSIGNED_BYTE, LLVertexBuffer::sTypeSize[TYPE_VERTEX], ptr);
+#endif
+		}
+		if (data_mask & MAP_VERTEX)
+		{
+			S32 loc = TYPE_VERTEX;
+			void* ptr = (void*)(base + mOffsets[TYPE_VERTEX]);
+			glVertexAttribPointerARB(loc, 3,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_VERTEX], ptr);
+		}	
+	}	
+	else
 	{
-		if (data_mask & MAP_TEXTURE_INDEX)
+		if (data_mask & MAP_NORMAL)
 		{
-			glVertexPointer(4,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_VERTEX], (void*)(base + 0));
+			glNormalPointer(GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_NORMAL], (void*)(base + mOffsets[TYPE_NORMAL]));
 		}
-		else
+		if (data_mask & MAP_TEXCOORD3)
 		{
-			glVertexPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_VERTEX], (void*)(base + 0));
+			glClientActiveTextureARB(GL_TEXTURE3_ARB);
+			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD3], (void*)(base + mOffsets[TYPE_TEXCOORD3]));
+			glClientActiveTextureARB(GL_TEXTURE0_ARB);
 		}
+		if (data_mask & MAP_TEXCOORD2)
+		{
+			glClientActiveTextureARB(GL_TEXTURE2_ARB);
+			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD2], (void*)(base + mOffsets[TYPE_TEXCOORD2]));
+			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+		}
+		if (data_mask & MAP_TEXCOORD1)
+		{
+			glClientActiveTextureARB(GL_TEXTURE1_ARB);
+			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD1], (void*)(base + mOffsets[TYPE_TEXCOORD1]));
+			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+		}
+		if (data_mask & MAP_BINORMAL)
+		{
+			glClientActiveTextureARB(GL_TEXTURE2_ARB);
+			glTexCoordPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_BINORMAL], (void*)(base + mOffsets[TYPE_BINORMAL]));
+			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+		}
+		if (data_mask & MAP_TEXCOORD0)
+		{
+			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD0], (void*)(base + mOffsets[TYPE_TEXCOORD0]));
+		}
+		if (data_mask & MAP_COLOR)
+		{
+			glColorPointer(4, GL_UNSIGNED_BYTE, LLVertexBuffer::sTypeSize[TYPE_COLOR], (void*)(base + mOffsets[TYPE_COLOR]));
+		}
+		if (data_mask & MAP_VERTEX)
+		{
+			glVertexPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_VERTEX], (void*)(base + 0));
+		}	
 	}
 
 	llglassertok();
 }
 
+LLVertexBuffer::MappedRegion::MappedRegion(S32 type, S32 index, S32 count)
+: mType(type), mIndex(index), mCount(count)
+{ 
+	llassert(mType == LLVertexBuffer::TYPE_INDEX || 
+			mType < LLVertexBuffer::TYPE_TEXTURE_INDEX);
+}	
+
+
diff --git a/indra/llrender/llvertexbuffer.h b/indra/llrender/llvertexbuffer.h
index 578cec3885..d859199663 100644
--- a/indra/llrender/llvertexbuffer.h
+++ b/indra/llrender/llvertexbuffer.h
@@ -38,6 +38,8 @@
 #include <vector>
 #include <list>
 
+#define LL_MAX_VERTEX_ATTRIB_LOCATION 64
+
 //============================================================================
 // NOTES
 // Threading:
@@ -49,25 +51,37 @@
 
 //============================================================================
 // gl name pools for dynamic and streaming buffers
-
-class LLVBOPool : public LLGLNamePool
+class LLVBOPool
 {
-protected:
-	virtual GLuint allocateName()
-	{
-		GLuint name;
-		stop_glerror();
-		glGenBuffersARB(1, &name);
-		stop_glerror();
-		return name;
-	}
+public:
+	static U32 sBytesPooled;
+	
+	LLVBOPool(U32 vboUsage, U32 vboType)
+		: mUsage(vboUsage)
+		, mType(vboType)
+	{}
+
+	const U32 mUsage;
+	const U32 mType;
 
-	virtual void releaseName(GLuint name)
+	//size MUST be a power of 2
+	volatile U8* allocate(U32& name, U32 size);
+	
+	//size MUST be the size provided to allocate that returned the given name
+	void release(U32 name, volatile U8* buffer, U32 size);
+	
+	//destroy all records in mFreeList
+	void cleanup();
+
+	class Record
 	{
-		stop_glerror();
-		glDeleteBuffersARB(1, &name);
-		stop_glerror();
-	}
+	public:
+		U32 mGLName;
+		volatile U8* mClientData;
+	};
+
+	typedef std::list<Record> record_list_t;
+	std::vector<record_list_t> mFreeList;
 };
 
 class LLGLFence
@@ -79,7 +93,7 @@ public:
 
 //============================================================================
 // base class 
-class LLPrivateMemoryPool ;
+class LLPrivateMemoryPool;
 class LLVertexBuffer : public LLRefCount
 {
 public:
@@ -90,12 +104,11 @@ public:
 		S32 mIndex;
 		S32 mCount;
 		
-		MappedRegion(S32 type, S32 index, S32 count)
-			: mType(type), mIndex(index), mCount(count)
-		{ }	
+		MappedRegion(S32 type, S32 index, S32 count);
 	};
 
 	LLVertexBuffer(const LLVertexBuffer& rhs)
+		: mUsage(rhs.mUsage)
 	{
 		*this = rhs;
 	}
@@ -111,18 +124,17 @@ public:
 	static LLVBOPool sStreamIBOPool;
 	static LLVBOPool sDynamicIBOPool;
 
-	static S32	sWeight4Loc;
-
-	static BOOL	sUseStreamDraw;
-	static BOOL	sPreferStreamDraw;
+	static bool	sUseStreamDraw;
+	static bool sUseVAO;
+	static bool	sPreferStreamDraw;
 
 	static void initClass(bool use_vbo, bool no_vbo_mapping);
 	static void cleanupClass();
 	static void setupClientArrays(U32 data_mask);
 	static void drawArrays(U32 mode, const std::vector<LLVector3>& pos, const std::vector<LLVector3>& norm);
+	static void drawElements(U32 mode, const LLVector4a* pos, const LLVector2* tc, S32 num_indices, const U16* indicesp);
 
- 	static void clientCopy(F64 max_time = 0.005); //copy data from client to GL
-	static void unbind(); //unbind any bound vertex buffer
+ 	static void unbind(); //unbind any bound vertex buffer
 
 	//get the size of a vertex with the given typemask
 	static S32 calcVertexSize(const U32& typemask);
@@ -133,24 +145,29 @@ public:
 	static S32 calcOffsets(const U32& typemask, S32* offsets, S32 num_vertices);		
 
 	
+	//WARNING -- when updating these enums you MUST 
+	// 1 - update LLVertexBuffer::sTypeSize
+	// 2 - add a strider accessor
+	// 3 - modify LLVertexBuffer::setupVertexBuffer
+	// 4 - modify LLVertexBuffer::setupClientArray
+	// 5 - modify LLViewerShaderMgr::mReservedAttribs
+	// 6 - update LLVertexBuffer::setupVertexArray
 	enum {
-		TYPE_VERTEX,
+		TYPE_VERTEX = 0,
 		TYPE_NORMAL,
 		TYPE_TEXCOORD0,
 		TYPE_TEXCOORD1,
 		TYPE_TEXCOORD2,
 		TYPE_TEXCOORD3,
 		TYPE_COLOR,
-		// These use VertexAttribPointer and should possibly be made generic
+		TYPE_EMISSIVE,
 		TYPE_BINORMAL,
 		TYPE_WEIGHT,
 		TYPE_WEIGHT4,
 		TYPE_CLOTHWEIGHT,
-		TYPE_MAX,
-		TYPE_INDEX,
-		
-		//no actual additional data, but indicates position.w is texture index
 		TYPE_TEXTURE_INDEX,
+		TYPE_MAX,
+		TYPE_INDEX,		
 	};
 	enum {
 		MAP_VERTEX = (1<<TYPE_VERTEX),
@@ -160,6 +177,7 @@ public:
 		MAP_TEXCOORD2 = (1<<TYPE_TEXCOORD2),
 		MAP_TEXCOORD3 = (1<<TYPE_TEXCOORD3),
 		MAP_COLOR = (1<<TYPE_COLOR),
+		MAP_EMISSIVE = (1<<TYPE_EMISSIVE),
 		// These use VertexAttribPointer and should possibly be made generic
 		MAP_BINORMAL = (1<<TYPE_BINORMAL),
 		MAP_WEIGHT = (1<<TYPE_WEIGHT),
@@ -173,33 +191,35 @@ protected:
 
 	virtual ~LLVertexBuffer(); // use unref()
 
-	virtual void setupVertexBuffer(U32 data_mask) const; // pure virtual, called from mapBuffer()
+	virtual void setupVertexBuffer(U32 data_mask); // pure virtual, called from mapBuffer()
+	void setupVertexArray();
 	
-	void	genBuffer();
-	void	genIndices();
+	void	genBuffer(U32 size);
+	void	genIndices(U32 size);
+	bool	bindGLBuffer(bool force_bind = false);
+	bool	bindGLIndices(bool force_bind = false);
+	bool	bindGLArray();
 	void	releaseBuffer();
 	void	releaseIndices();
-	void	createGLBuffer();
-	void	createGLIndices();
+	void	createGLBuffer(U32 size);
+	void	createGLIndices(U32 size);
 	void 	destroyGLBuffer();
 	void 	destroyGLIndices();
 	void	updateNumVerts(S32 nverts);
 	void	updateNumIndices(S32 nindices); 
-	virtual BOOL	useVBOs() const;
-	void	unmapBuffer(S32 type);
-	void freeClientBuffer() ;
-	void allocateClientVertexBuffer() ;
-	void allocateClientIndexBuffer() ;
-
+	bool	useVBOs() const;
+	void	unmapBuffer();
+		
 public:
 	LLVertexBuffer(U32 typemask, S32 usage);
 	
 	// map for data access
-	U8*		mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range);
-	U8*		mapIndexBuffer(S32 index, S32 count, bool map_range);
+	volatile U8*		mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range);
+	volatile U8*		mapIndexBuffer(S32 index, S32 count, bool map_range);
 
 	// set for rendering
-	virtual void	setBuffer(U32 data_mask, S32 type = -1); 	// calls  setupVertexBuffer() if data_mask is not 0
+	virtual void	setBuffer(U32 data_mask); 	// calls  setupVertexBuffer() if data_mask is not 0
+	void flush(); //flush pending data to GL memory
 	// allocate buffer
 	void	allocateBuffer(S32 nverts, S32 nindices, bool create);
 	virtual void resizeBuffer(S32 newnverts, S32 newnindices);
@@ -212,33 +232,35 @@ public:
 	//   setVertsNorms(verts, norms);
 	//   vb->unmapBuffer();
 	bool getVertexStrider(LLStrider<LLVector3>& strider, S32 index=0, S32 count = -1, bool map_range = false);
+	bool getVertexStrider(LLStrider<LLVector4a>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getIndexStrider(LLStrider<U16>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getTexCoord0Strider(LLStrider<LLVector2>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getTexCoord1Strider(LLStrider<LLVector2>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getNormalStrider(LLStrider<LLVector3>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getBinormalStrider(LLStrider<LLVector3>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getColorStrider(LLStrider<LLColor4U>& strider, S32 index=0, S32 count = -1, bool map_range = false);
+	bool getEmissiveStrider(LLStrider<LLColor4U>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getWeightStrider(LLStrider<F32>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getWeight4Strider(LLStrider<LLVector4>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	bool getClothWeightStrider(LLStrider<LLVector4>& strider, S32 index=0, S32 count = -1, bool map_range = false);
 	
-	BOOL isEmpty() const					{ return mEmpty; }
-	BOOL isLocked() const					{ return mVertexLocked || mIndexLocked; }
+
+	bool isEmpty() const					{ return mEmpty; }
+	bool isLocked() const					{ return mVertexLocked || mIndexLocked; }
 	S32 getNumVerts() const					{ return mNumVerts; }
 	S32 getNumIndices() const				{ return mNumIndices; }
-	S32 getRequestedVerts() const			{ return mRequestedNumVerts; }
-	S32 getRequestedIndices() const			{ return mRequestedNumIndices; }
-
-	U8* getIndicesPointer() const			{ return useVBOs() ? (U8*) mAlignedIndexOffset : mMappedIndexData; }
-	U8* getVerticesPointer() const			{ return useVBOs() ? (U8*) mAlignedOffset : mMappedData; }
+	
+	volatile U8* getIndicesPointer() const			{ return useVBOs() ? (U8*) mAlignedIndexOffset : mMappedIndexData; }
+	volatile U8* getVerticesPointer() const			{ return useVBOs() ? (U8*) mAlignedOffset : mMappedData; }
 	U32 getTypeMask() const					{ return mTypeMask; }
 	bool hasDataType(S32 type) const		{ return ((1 << type) & getTypeMask()); }
 	S32 getSize() const;
-	S32 getIndicesSize() const				{ return mNumIndices * sizeof(U16); }
-	U8* getMappedData() const				{ return mMappedData; }
-	U8* getMappedIndices() const			{ return mMappedIndexData; }
+	S32 getIndicesSize() const				{ return mIndicesSize; }
+	volatile U8* getMappedData() const				{ return mMappedData; }
+	volatile U8* getMappedIndices() const			{ return mMappedIndexData; }
 	S32 getOffset(S32 type) const			{ return mOffsets[type]; }
 	S32 getUsage() const					{ return mUsage; }
+	bool isWriteable() const				{ return (mMappable || mUsage == GL_STREAM_DRAW_ARB) ? true : false; }
 
 	void draw(U32 mode, U32 count, U32 indices_offset) const;
 	void drawArrays(U32 mode, U32 offset, U32 count) const;
@@ -252,25 +274,31 @@ public:
 protected:	
 	S32		mNumVerts;		// Number of vertices allocated
 	S32		mNumIndices;	// Number of indices allocated
-	S32		mRequestedNumVerts;  // Number of vertices requested
-	S32		mRequestedNumIndices;  // Number of indices requested
-
+	
 	ptrdiff_t mAlignedOffset;
 	ptrdiff_t mAlignedIndexOffset;
 	S32		mSize;
+	S32		mIndicesSize;
 	U32		mTypeMask;
-	S32		mUsage;			// GL usage
+
+	const S32		mUsage;			// GL usage
+	
 	U32		mGLBuffer;		// GL VBO handle
 	U32		mGLIndices;		// GL IBO handle
-	U8*		mMappedData;	// pointer to currently mapped data (NULL if unmapped)
-	U8*		mMappedIndexData;	// pointer to currently mapped indices (NULL if unmapped)
-	BOOL	mVertexLocked;			// if TRUE, vertex buffer is being or has been written to in client memory
-	BOOL	mIndexLocked;			// if TRUE, index buffer is being or has been written to in client memory
-	BOOL	mFinal;			// if TRUE, buffer can not be mapped again
-	BOOL	mFilthy;		// if TRUE, entire buffer must be copied (used to prevent redundant dirty flags)
-	BOOL	mEmpty;			// if TRUE, client buffer is empty (or NULL). Old values have been discarded.	
-	BOOL	mResized;		// if TRUE, client buffer has been resized and GL buffer has not
-	BOOL	mDynamicSize;	// if TRUE, buffer has been resized at least once (and should be padded)
+	U32		mGLArray;		// GL VAO handle
+	
+	volatile U8* mMappedData;	// pointer to currently mapped data (NULL if unmapped)
+	volatile U8* mMappedIndexData;	// pointer to currently mapped indices (NULL if unmapped)
+
+	U32		mMappedDataUsingVBOs : 1;
+	U32		mMappedIndexDataUsingVBOs : 1;
+	U32		mVertexLocked : 1;			// if true, vertex buffer is being or has been written to in client memory
+	U32		mIndexLocked : 1;			// if true, index buffer is being or has been written to in client memory
+	U32		mFinal : 1;			// if true, buffer can not be mapped again
+	U32		mEmpty : 1;			// if true, client buffer is empty (or NULL). Old values have been discarded.	
+	
+	mutable bool	mMappable;     // if true, use memory mapping to upload data (otherwise doublebuffer and use glBufferSubData)
+
 	S32		mOffsets[TYPE_MAX];
 
 	std::vector<MappedRegion> mMappedVertexRegions;
@@ -281,26 +309,27 @@ protected:
 	void placeFence() const;
 	void waitFence() const;
 
+	static S32 determineUsage(S32 usage);
 
 private:
-	static LLPrivateMemoryPool* sPrivatePoolp ;
+	static LLPrivateMemoryPool* sPrivatePoolp;
 
 public:
 	static S32 sCount;
 	static S32 sGLCount;
 	static S32 sMappedCount;
-	static BOOL sMapped;
-	static std::vector<U32> sDeleteList;
+	static bool sMapped;
 	typedef std::list<LLVertexBuffer*> buffer_list_t;
 		
-	static BOOL sDisableVBOMapping; //disable glMapBufferARB
-	static BOOL sEnableVBOs;
+	static bool sDisableVBOMapping; //disable glMapBufferARB
+	static bool sEnableVBOs;
 	static S32 sTypeSize[TYPE_MAX];
 	static U32 sGLMode[LLRender::NUM_MODES];
 	static U32 sGLRenderBuffer;
+	static U32 sGLRenderArray;
 	static U32 sGLRenderIndices;
-	static BOOL sVBOActive;
-	static BOOL sIBOActive;
+	static bool sVBOActive;
+	static bool sIBOActive;
 	static U32 sLastMask;
 	static U32 sAllocatedBytes;
 	static U32 sBindCount;