Merge branch 'master' v 6.6.1 into DRTVWR-528

28 files changed, 1571 insertions, 2101 deletions

diff --git a/indra/llrender/llcubemap.cpp b/indra/llrender/llcubemap.cpp
index 5947bca670..834084674e 100644
--- a/indra/llrender/llcubemap.cpp
+++ b/indra/llrender/llcubemap.cpp
@@ -47,7 +47,6 @@ bool LLCubeMap::sUseCubeMaps = true;
 
 LLCubeMap::LLCubeMap(bool init_as_srgb)
 	: mTextureStage(0),
-	  mTextureCoordStage(0),
 	  mMatrixStage(0),
 	  mIssRGB(init_as_srgb)
 {
@@ -150,6 +149,7 @@ void LLCubeMap::initRawData(const std::vector<LLPointer<LLImageRaw> >& rawimages
 
 void LLCubeMap::initGLData()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	for (int i = 0; i < 6; i++)
 	{
 		mImages[i]->setSubImage(mRawImages[i], 0, 0, RESOLUTION, RESOLUTION);
@@ -179,7 +179,6 @@ void LLCubeMap::bind()
 void LLCubeMap::enable(S32 stage)
 {
 	enableTexture(stage);
-	enableTextureCoords(stage);
 }
 
 void LLCubeMap::enableTexture(S32 stage)
@@ -191,35 +190,9 @@ void LLCubeMap::enableTexture(S32 stage)
 	}
 }
 
-void LLCubeMap::enableTextureCoords(S32 stage)
-{
-	mTextureCoordStage = stage;
-	if (!LLGLSLShader::sNoFixedFunction && gGLManager.mHasCubeMap && stage >= 0 && LLCubeMap::sUseCubeMaps)
-	{
-		if (stage > 0)
-		{
-			gGL.getTexUnit(stage)->activate();
-		}
-		
-		glEnable(GL_TEXTURE_GEN_R);
-		glEnable(GL_TEXTURE_GEN_S);
-		glEnable(GL_TEXTURE_GEN_T);
-
-		glTexGeni(GL_S, GL_TEXTURE_GEN_MODE, GL_REFLECTION_MAP);
-		glTexGeni(GL_T, GL_TEXTURE_GEN_MODE, GL_REFLECTION_MAP);
-		glTexGeni(GL_R, GL_TEXTURE_GEN_MODE, GL_REFLECTION_MAP);
-		
-		if (stage > 0)
-		{
-			gGL.getTexUnit(0)->activate();
-		}
-	}
-}
-
 void LLCubeMap::disable(void)
 {
 	disableTexture();
-	disableTextureCoords();
 }
 
 void LLCubeMap::disableTexture(void)
@@ -234,24 +207,6 @@ void LLCubeMap::disableTexture(void)
 	}
 }
 
-void LLCubeMap::disableTextureCoords(void)
-{
-	if (!LLGLSLShader::sNoFixedFunction && gGLManager.mHasCubeMap && mTextureCoordStage >= 0 && LLCubeMap::sUseCubeMaps)
-	{
-		if (mTextureCoordStage > 0)
-		{
-			gGL.getTexUnit(mTextureCoordStage)->activate();
-		}
-		glDisable(GL_TEXTURE_GEN_S);
-		glDisable(GL_TEXTURE_GEN_T);
-		glDisable(GL_TEXTURE_GEN_R);
-		if (mTextureCoordStage > 0)
-		{
-			gGL.getTexUnit(0)->activate();
-		}
-	}
-}
-
 void LLCubeMap::setMatrix(S32 stage)
 {
 	mMatrixStage = stage;
@@ -453,6 +408,7 @@ BOOL LLCubeMap::project(F32& v_min, F32& v_max, F32& h_min, F32& h_max,
 
 void LLCubeMap::paintIn(LLVector3 dir[4], const LLColor4U& col)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	F32 v_min, v_max, h_min, h_max;
 	LLVector3 center = dir[0] + dir[1] + dir[2] + dir[3];
 	center.normVec();
diff --git a/indra/llrender/llcubemap.h b/indra/llrender/llcubemap.h
index 95b6d12099..a01636d8d4 100644
--- a/indra/llrender/llcubemap.h
+++ b/indra/llrender/llcubemap.h
@@ -48,12 +48,10 @@ public:
 	void enable(S32 stage);
 	
 	void enableTexture(S32 stage);
-	void enableTextureCoords(S32 stage);
 	S32	 getStage(void) { return mTextureStage; }
 	
 	void disable(void);
 	void disableTexture(void);
-	void disableTextureCoords(void);
 	void setMatrix(S32 stage);
 	void restoreMatrix();
 	void setReflection (void);
@@ -80,7 +78,6 @@ protected:
 	LLPointer<LLImageGL> mImages[6];
 	LLPointer<LLImageRaw> mRawImages[6];
 	S32 mTextureStage;
-	S32 mTextureCoordStage;
 	S32 mMatrixStage;
 };
 
diff --git a/indra/llrender/llfontbitmapcache.cpp b/indra/llrender/llfontbitmapcache.cpp
index f128636ab2..c71e24c83a 100644
--- a/indra/llrender/llfontbitmapcache.cpp
+++ b/indra/llrender/llfontbitmapcache.cpp
@@ -30,8 +30,7 @@
 #include "llfontbitmapcache.h"
 
 LLFontBitmapCache::LLFontBitmapCache()
-:	LLTrace::MemTrackable<LLFontBitmapCache>("LLFontBitmapCache"),
-	mNumComponents(0),
+:	mNumComponents(0),
 	mBitmapWidth(0),
 	mBitmapHeight(0),
 	mBitmapNum(-1),
@@ -124,9 +123,6 @@ BOOL LLFontBitmapCache::nextOpenPos(S32 width, S32 &pos_x, S32 &pos_y, S32& bitm
 			image_gl->createGLTexture(0, image_raw);
 			gGL.getTexUnit(0)->bind(image_gl);
 			image_gl->setFilteringOption(LLTexUnit::TFO_POINT); // was setMipFilterNearest(TRUE, TRUE);
-
-			claimMem(image_raw);
-			claimMem(image_gl);
 		}
 		else
 		{
@@ -156,20 +152,8 @@ void LLFontBitmapCache::destroyGL()
 
 void LLFontBitmapCache::reset()
 {
-	for (std::vector<LLPointer<LLImageRaw> >::iterator it = mImageRawVec.begin(), end_it = mImageRawVec.end();
-		it != end_it;
-		++it)
-	{
-		disclaimMem(**it);
-	}
 	mImageRawVec.clear();
 
-	for (std::vector<LLPointer<LLImageGL> >::iterator it = mImageGLVec.begin(), end_it = mImageGLVec.end();
-		it != end_it;
-		++it)
-	{
-		disclaimMem(**it);
-	}
 	mImageGLVec.clear();
 	
 	mBitmapWidth = 0;
diff --git a/indra/llrender/llfontbitmapcache.h b/indra/llrender/llfontbitmapcache.h
index 75df3a94a7..7de3a6b56f 100644
--- a/indra/llrender/llfontbitmapcache.h
+++ b/indra/llrender/llfontbitmapcache.h
@@ -32,7 +32,7 @@
 
 // Maintain a collection of bitmaps containing rendered glyphs.
 // Generalizes the single-bitmap logic from LLFontFreetype and LLFontGL.
-class LLFontBitmapCache : public LLTrace::MemTrackable<LLFontBitmapCache>
+class LLFontBitmapCache
 {
 public:
 	LLFontBitmapCache();
diff --git a/indra/llrender/llfontfreetype.cpp b/indra/llrender/llfontfreetype.cpp
index c41730ebaa..e964d1586f 100644
--- a/indra/llrender/llfontfreetype.cpp
+++ b/indra/llrender/llfontfreetype.cpp
@@ -104,8 +104,7 @@ LLFontGlyphInfo::LLFontGlyphInfo(U32 index)
 }
 
 LLFontFreetype::LLFontFreetype()
-:	LLTrace::MemTrackable<LLFontFreetype>("LLFontFreetype"),
-	mFontBitmapCachep(new LLFontBitmapCache),
+:	mFontBitmapCachep(new LLFontBitmapCache),
 	mAscender(0.f),
 	mDescender(0.f),
 	mLineHeight(0.f),
@@ -222,8 +221,6 @@ BOOL LLFontFreetype::loadFace(const std::string& filename, F32 point_size, F32 v
 	S32 max_char_height = ll_round(0.5f + (y_max - y_min));
 
 	mFontBitmapCachep->init(components, max_char_width, max_char_height);
-	claimMem(mFontBitmapCachep);
-
 
 	if (!mFTFace->charmap)
 	{
@@ -238,7 +235,6 @@ BOOL LLFontFreetype::loadFace(const std::string& filename, F32 point_size, F32 v
 	}
 
 	mName = filename;
-	claimMem(mName);
 	mPointSize = point_size;
 
 	mStyle = LLFontGL::NORMAL;
@@ -460,6 +456,7 @@ LLFontGlyphInfo* LLFontFreetype::addGlyph(llwchar wch) const
 
 LLFontGlyphInfo* LLFontFreetype::addGlyphFromFont(const LLFontFreetype *fontp, llwchar wch, U32 glyph_index) const
 {
+    LL_PROFILE_ZONE_SCOPED;
 	if (mFTFace == NULL)
 		return NULL;
 
@@ -585,7 +582,6 @@ void LLFontFreetype::insertGlyphInfo(llwchar wch, LLFontGlyphInfo* gi) const
 	}
 	else
 	{
-		claimMem(gi);
 		mCharGlyphInfoMap[wch] = gi;
 	}
 }
@@ -631,11 +627,9 @@ void LLFontFreetype::resetBitmapCache()
 		it != end_it;
 		++it)
 	{
-		disclaimMem(it->second);
 		delete it->second;
 	}
 	mCharGlyphInfoMap.clear();
-	disclaimMem(mFontBitmapCachep);
 	mFontBitmapCachep->reset();
 
 	// Adding default glyph is skipped for fallback fonts here as well as in loadFace(). 
diff --git a/indra/llrender/llfontfreetype.h b/indra/llrender/llfontfreetype.h
index 1afe84e770..f61f169987 100644
--- a/indra/llrender/llfontfreetype.h
+++ b/indra/llrender/llfontfreetype.h
@@ -76,7 +76,7 @@ struct LLFontGlyphInfo
 
 extern LLFontManager *gFontManagerp;
 
-class LLFontFreetype : public LLRefCount, public LLTrace::MemTrackable<LLFontFreetype>
+class LLFontFreetype : public LLRefCount
 {
 public:
 	LLFontFreetype();
diff --git a/indra/llrender/llfontgl.cpp b/indra/llrender/llfontgl.cpp
index 86a4c35e6d..1bf061bc8d 100644
--- a/indra/llrender/llfontgl.cpp
+++ b/indra/llrender/llfontgl.cpp
@@ -109,8 +109,6 @@ S32 LLFontGL::getNumFaces(const std::string& filename)
 	return mFontFreetype->getNumFaces(filename);
 }
 
-static LLTrace::BlockTimerStatHandle 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
 {
@@ -147,7 +145,7 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, const LLRectf& rec
 S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, const LLColor4 &color, HAlign halign, VAlign valign, U8 style, 
 					 ShadowType shadow, S32 max_chars, S32 max_pixels, F32* right_x, BOOL use_ellipses) const
 {
-	LL_RECORD_BLOCK_TIME(FTM_RENDER_FONTS);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_UI;
 
 	if(!sDisplayFont) //do not display texts
 	{
@@ -547,9 +545,19 @@ F32 LLFontGL::getWidthF32(const llwchar* wchars, S32 begin_offset, S32 max_chars
 	return cur_x / sScaleX;
 }
 
+void LLFontGL::generateASCIIglyphs()
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_UI
+    for (U32 i = 32; (i < 127); i++)
+    {
+        mFontFreetype->getGlyphInfo(i);
+    }
+}
+
 // Returns the max number of complete characters from text (up to max_chars) that can be drawn in max_pixels
 S32 LLFontGL::maxDrawableChars(const llwchar* wchars, F32 max_pixels, S32 max_chars, EWordWrapStyle end_on_word_boundary) const
 {
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_UI
 	if (!wchars || !wchars[0] || max_chars == 0)
 	{
 		return 0;
@@ -829,6 +837,8 @@ void LLFontGL::initClass(F32 screen_dpi, F32 x_scale, F32 y_scale, const std::st
 	{
 		sFontRegistry->reset();
 	}
+
+	LLFontGL::loadDefaultFonts();
 }
 
 // Force standard fonts to get generated up front.
@@ -838,6 +848,7 @@ void LLFontGL::initClass(F32 screen_dpi, F32 x_scale, F32 y_scale, const std::st
 // static
 bool LLFontGL::loadDefaultFonts()
 {
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_UI
 	bool succ = true;
 	succ &= (NULL != getFontSansSerifSmall());
 	succ &= (NULL != getFontSansSerif());
@@ -845,10 +856,18 @@ bool LLFontGL::loadDefaultFonts()
 	succ &= (NULL != getFontSansSerifHuge());
 	succ &= (NULL != getFontSansSerifBold());
 	succ &= (NULL != getFontMonospace());
-	succ &= (NULL != getFontExtChar());
 	return succ;
 }
 
+void LLFontGL::loadCommonFonts()
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_UI
+    getFont(LLFontDescriptor("SansSerif", "Small", BOLD));
+    getFont(LLFontDescriptor("SansSerif", "Large", BOLD));
+    getFont(LLFontDescriptor("SansSerif", "Huge", BOLD));
+    getFont(LLFontDescriptor("Monospace", "Medium", 0));
+}
+
 // static
 void LLFontGL::destroyDefaultFonts()
 {
@@ -1015,7 +1034,7 @@ LLFontGL* LLFontGL::getFontSansSerifBig()
 //static 
 LLFontGL* LLFontGL::getFontSansSerifHuge()
 {
-	static LLFontGL* fontp = getFont(LLFontDescriptor("SansSerif","Large",0));
+	static LLFontGL* fontp = getFont(LLFontDescriptor("SansSerif","Huge",0));
 	return fontp;
 }
 
@@ -1026,12 +1045,6 @@ LLFontGL* LLFontGL::getFontSansSerifBold()
 	return fontp;
 }
 
-//static
-LLFontGL* LLFontGL::getFontExtChar()
-{
-	return getFontSansSerif();
-}
-
 //static 
 LLFontGL* LLFontGL::getFont(const LLFontDescriptor& desc)
 {
diff --git a/indra/llrender/llfontgl.h b/indra/llrender/llfontgl.h
index 10891faed9..3b58a37d33 100644
--- a/indra/llrender/llfontgl.h
+++ b/indra/llrender/llfontgl.h
@@ -160,12 +160,15 @@ public:
 
 	const LLFontDescriptor& getFontDesc() const;
 
+	void generateASCIIglyphs();
+
 
 	static void initClass(F32 screen_dpi, F32 x_scale, F32 y_scale, const std::string& app_dir, bool create_gl_textures = true);
 
 	// Load sans-serif, sans-serif-small, etc.
 	// Slow, requires multiple seconds to load fonts.
 	static bool loadDefaultFonts();
+    static void loadCommonFonts();
 	static void	destroyDefaultFonts();
 	static void destroyAllGL();
 
@@ -190,7 +193,6 @@ public:
 	static LLFontGL* getFontSansSerifBig();
 	static LLFontGL* getFontSansSerifHuge();
 	static LLFontGL* getFontSansSerifBold();
-	static LLFontGL* getFontExtChar();
 	static LLFontGL* getFont(const LLFontDescriptor& desc);
 	// Use with legacy names like "SANSSERIF_SMALL" or "OCRA"
 	static LLFontGL* getFontByName(const std::string& name);
diff --git a/indra/llrender/llfontregistry.cpp b/indra/llrender/llfontregistry.cpp
index 33a33af160..bc1a2f8887 100644
--- a/indra/llrender/llfontregistry.cpp
+++ b/indra/llrender/llfontregistry.cpp
@@ -597,6 +597,11 @@ LLFontGL *LLFontRegistry::getFont(const LLFontDescriptor& desc)
 					<<" style=[" << ((S32) desc.getStyle()) << "]"
 					<< " size=[" << desc.getSize() << "]" << LL_ENDL;
 		}
+		else
+		{
+			//generate glyphs for ASCII chars to avoid stalls later
+			fontp->generateASCIIglyphs();
+		}
 		return fontp;
 	}
 }
diff --git a/indra/llrender/llgl.cpp b/indra/llrender/llgl.cpp
index 43fedeca64..18168d1c3f 100644
--- a/indra/llrender/llgl.cpp
+++ b/indra/llrender/llgl.cpp
@@ -62,6 +62,7 @@
 BOOL gDebugSession = FALSE;
 BOOL gClothRipple = FALSE;
 BOOL gHeadlessClient = FALSE;
+BOOL gNonInteractive = FALSE;
 BOOL gGLActive = FALSE;
 BOOL gGLDebugLoggingEnabled = TRUE;
 
@@ -87,24 +88,32 @@ void APIENTRY gl_debug_callback(GLenum source,
 {
 	if (gGLDebugLoggingEnabled)
 	{
-	if (severity == GL_DEBUG_SEVERITY_HIGH_ARB)
-	{
-		LL_WARNS() << "----- GL ERROR --------" << LL_ENDL;
-	}
-	else
-	{
-		LL_WARNS() << "----- GL WARNING -------" << LL_ENDL;
-	}
-	LL_WARNS() << "Type: " << std::hex << type << LL_ENDL;
-	LL_WARNS() << "ID: " << std::hex << id << LL_ENDL;
-	LL_WARNS() << "Severity: " << std::hex << severity << LL_ENDL;
-	LL_WARNS() << "Message: " << message << LL_ENDL;
-	LL_WARNS() << "-----------------------" << LL_ENDL;
-	if (severity == GL_DEBUG_SEVERITY_HIGH_ARB)
-	{
-		LL_ERRS() << "Halting on GL Error" << LL_ENDL;
-	}
-}
+
+        if (severity != GL_DEBUG_SEVERITY_HIGH_ARB &&
+            severity != GL_DEBUG_SEVERITY_MEDIUM_ARB &&
+            severity != GL_DEBUG_SEVERITY_LOW_ARB)
+        { //suppress out-of-spec messages sent by nvidia driver (mostly vertexbuffer hints)
+            return;
+        }
+
+	    if (severity == GL_DEBUG_SEVERITY_HIGH_ARB)
+	    {
+		    LL_WARNS() << "----- GL ERROR --------" << LL_ENDL;
+	    }
+	    else
+	    {
+		    LL_WARNS() << "----- GL WARNING -------" << LL_ENDL;
+	    }
+	    LL_WARNS() << "Type: " << std::hex << type << LL_ENDL;
+	    LL_WARNS() << "ID: " << std::hex << id << LL_ENDL;
+	    LL_WARNS() << "Severity: " << std::hex << severity << LL_ENDL;
+	    LL_WARNS() << "Message: " << message << LL_ENDL;
+	    LL_WARNS() << "-----------------------" << LL_ENDL;
+	    if (severity == GL_DEBUG_SEVERITY_HIGH_ARB)
+	    {
+		    LL_ERRS() << "Halting on GL Error" << LL_ENDL;
+	    }
+    }
 }
 #endif
 
@@ -152,7 +161,6 @@ LLMatrix4 gGLObliqueProjectionInverse;
 std::list<LLGLUpdate*> LLGLUpdate::sGLQ;
 
 #if (LL_WINDOWS || LL_LINUX)  && !LL_MESA_HEADLESS
-// ATI prototypes
 
 #if LL_WINDOWS
 PFNGLGETSTRINGIPROC glGetStringi = NULL;
@@ -199,21 +207,6 @@ PFNGLGETSYNCIVPROC				glGetSynciv = NULL;
 PFNGLBUFFERPARAMETERIAPPLEPROC	glBufferParameteriAPPLE = NULL;
 PFNGLFLUSHMAPPEDBUFFERRANGEAPPLEPROC glFlushMappedBufferRangeAPPLE = NULL;
 
-// vertex object prototypes
-PFNGLNEWOBJECTBUFFERATIPROC			glNewObjectBufferATI = NULL;
-PFNGLISOBJECTBUFFERATIPROC			glIsObjectBufferATI = NULL;
-PFNGLUPDATEOBJECTBUFFERATIPROC		glUpdateObjectBufferATI = NULL;
-PFNGLGETOBJECTBUFFERFVATIPROC		glGetObjectBufferfvATI = NULL;
-PFNGLGETOBJECTBUFFERIVATIPROC		glGetObjectBufferivATI = NULL;
-PFNGLFREEOBJECTBUFFERATIPROC		glFreeObjectBufferATI = NULL;
-PFNGLARRAYOBJECTATIPROC				glArrayObjectATI = NULL;
-PFNGLVERTEXATTRIBARRAYOBJECTATIPROC	glVertexAttribArrayObjectATI = NULL;
-PFNGLGETARRAYOBJECTFVATIPROC		glGetArrayObjectfvATI = NULL;
-PFNGLGETARRAYOBJECTIVATIPROC		glGetArrayObjectivATI = NULL;
-PFNGLVARIANTARRAYOBJECTATIPROC		glVariantObjectArrayATI = NULL;
-PFNGLGETVARIANTARRAYOBJECTFVATIPROC	glGetVariantArrayObjectfvATI = NULL;
-PFNGLGETVARIANTARRAYOBJECTIVATIPROC	glGetVariantArrayObjectivATI = NULL;
-
 // GL_ARB_occlusion_query
 PFNGLGENQUERIESARBPROC glGenQueriesARB = NULL;
 PFNGLDELETEQUERIESARBPROC glDeleteQueriesARB = NULL;
@@ -434,9 +427,6 @@ LLGLManager::LLGLManager() :
 	mHasMapBufferRange(FALSE),
 	mHasFlushBufferRange(FALSE),
 	mHasPBuffer(FALSE),
-	mHasShaderObjects(FALSE),
-	mHasVertexShader(FALSE),
-	mHasFragmentShader(FALSE),
 	mNumTextureImageUnits(0),
 	mHasOcclusionQuery(FALSE),
 	mHasTimerQuery(FALSE),
@@ -456,14 +446,9 @@ LLGLManager::LLGLManager() :
 	mHasCubeMap(FALSE),
 	mHasDebugOutput(FALSE),
 
-	mIsATI(FALSE),
+	mIsAMD(FALSE),
 	mIsNVIDIA(FALSE),
 	mIsIntel(FALSE),
-	mIsGF2or4MX(FALSE),
-	mIsGF3(FALSE),
-	mIsGFFX(FALSE),
-	mATIOffsetVerticalLines(FALSE),
-	mATIOldDriver(FALSE),
 #if LL_DARWIN
 	mIsMobileGF(FALSE),
 #endif
@@ -559,7 +544,7 @@ bool LLGLManager::initGL()
 		glGetIntegerv(GL_NUM_EXTENSIONS, &count);
 		for (GLint i = 0; i < count; ++i)
 		{
-			std::string ext((const char*) glGetStringi(GL_EXTENSIONS, i));
+			std::string ext = ll_safe_string((const char*) glGetStringi(GL_EXTENSIONS, i));
 			str << ext << " ";
 			LL_DEBUGS("GLExtensions") << ext << LL_ENDL;
 		}
@@ -622,59 +607,17 @@ bool LLGLManager::initGL()
 	
 	// Trailing space necessary to keep "nVidia Corpor_ati_on" cards
 	// from being recognized as ATI.
+    // NOTE: AMD has been pretty good about not breaking this check, do not rename without good reason
 	if (mGLVendor.substr(0,4) == "ATI ")
 	{
-		mGLVendorShort = "ATI";
+		mGLVendorShort = "AMD";
 		// *TODO: Fix this?
-		mIsATI = TRUE;
-
-#if LL_WINDOWS && !LL_MESA_HEADLESS
-		if (mDriverVersionRelease < 3842)
-		{
-			mATIOffsetVerticalLines = TRUE;
-		}
-#endif // LL_WINDOWS
-
-#if (LL_WINDOWS || LL_LINUX) && !LL_MESA_HEADLESS
-		// count any pre OpenGL 3.0 implementation as an old driver
-		if (mGLVersion < 3.f) 
-		{
-			mATIOldDriver = TRUE;
-		}
-#endif // (LL_WINDOWS || LL_LINUX) && !LL_MESA_HEADLESS
+		mIsAMD = TRUE;
 	}
 	else if (mGLVendor.find("NVIDIA ") != std::string::npos)
 	{
 		mGLVendorShort = "NVIDIA";
 		mIsNVIDIA = TRUE;
-		if (   mGLRenderer.find("GEFORCE4 MX") != std::string::npos
-			|| mGLRenderer.find("GEFORCE2") != std::string::npos
-			|| mGLRenderer.find("GEFORCE 2") != std::string::npos
-			|| mGLRenderer.find("GEFORCE4 460 GO") != std::string::npos
-			|| mGLRenderer.find("GEFORCE4 440 GO") != std::string::npos
-			|| mGLRenderer.find("GEFORCE4 420 GO") != std::string::npos)
-		{
-			mIsGF2or4MX = TRUE;
-		}
-		else if (mGLRenderer.find("GEFORCE FX") != std::string::npos
-				 || mGLRenderer.find("QUADRO FX") != std::string::npos
-				 || mGLRenderer.find("NV34") != std::string::npos)
-		{
-			mIsGFFX = TRUE;
-		}
-		else if(mGLRenderer.find("GEFORCE3") != std::string::npos)
-		{
-			mIsGF3 = TRUE;
-		}
-#if LL_DARWIN
-		else if ((mGLRenderer.find("9400M") != std::string::npos)
-			  || (mGLRenderer.find("9600M") != std::string::npos)
-			  || (mGLRenderer.find("9800M") != std::string::npos))
-		{
-			mIsMobileGF = TRUE;
-		}
-#endif
-
 	}
 	else if (mGLVendor.find("INTEL") != std::string::npos
 #if LL_LINUX
@@ -775,29 +718,27 @@ bool LLGLManager::initGL()
 
 	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);
-	}
+	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);
-		mNumTextureUnits = llmin(num_tex_units, (GLint)MAX_GL_TEXTURE_UNITS);
-		if (mIsIntel)
-		{
-			mNumTextureUnits = llmin(mNumTextureUnits, 2);
-		}
-	}
+    if (mHasMultitexture)
+    {
+        if (LLRender::sGLCoreProfile)
+        {
+            mNumTextureUnits = llmin(mNumTextureImageUnits, MAX_GL_TEXTURE_UNITS);
+        }
+        else
+        {
+            GLint num_tex_units;
+            glGetIntegerv(GL_MAX_TEXTURE_UNITS_ARB, &num_tex_units);
+            mNumTextureUnits = llmin(num_tex_units, (GLint)MAX_GL_TEXTURE_UNITS);
+            if (mIsIntel)
+            {
+                mNumTextureUnits = llmin(mNumTextureUnits, 2);
+            }
+        }
+    }
 	else
 	{
 		mHasRequirements = FALSE;
@@ -819,26 +760,9 @@ bool LLGLManager::initGL()
 
 	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
-		mHasTextureMultisample = FALSE;
-	}
-
-
 	if (mIsIntel && mGLVersion <= 3.f)
 	{ //never try to use framebuffer objects on older intel drivers (crashy)
 		mHasFramebufferObject = FALSE;
@@ -975,9 +899,9 @@ void LLGLManager::asLLSD(LLSD& info)
 	info["has_map_buffer_range"] = mHasMapBufferRange;
 	info["has_flush_buffer_range"] = mHasFlushBufferRange;
 	info["has_pbuffer"] = mHasPBuffer;
-	info["has_shader_objects"] = mHasShaderObjects;
-	info["has_vertex_shader"] = mHasVertexShader;
-	info["has_fragment_shader"] = mHasFragmentShader;
+    info["has_shader_objects"] = std::string("Assumed TRUE");   // was mHasShaderObjects;
+	info["has_vertex_shader"] = std::string("Assumed TRUE");    // was mHasVertexShader;
+	info["has_fragment_shader"] = std::string("Assumed TRUE");  // was mHasFragmentShader;
 	info["num_texture_image_units"] =  mNumTextureImageUnits;
 	info["has_occlusion_query"] = mHasOcclusionQuery;
 	info["has_timer_query"] = mHasTimerQuery;
@@ -1003,14 +927,9 @@ void LLGLManager::asLLSD(LLSD& info)
     info["has_texture_srgb_decode"] = mHasTexturesRGBDecode;
 
 	// Vendor-specific extensions
-	info["is_ati"] = mIsATI;
+	info["is_ati"] = mIsAMD;  // note, do not rename is_ati to is_amd without coordinating with DW
 	info["is_nvidia"] = mIsNVIDIA;
 	info["is_intel"] = mIsIntel;
-	info["is_gf2or4mx"] = mIsGF2or4MX;
-	info["is_gf3"] = mIsGF3;
-	info["is_gf_gfx"] = mIsGFFX;
-	info["ati_offset_vertical_lines"] = mATIOffsetVerticalLines;
-	info["ati_old_driver"] = mATIOldDriver;
 
 	// Other fields
 	info["has_requirements"] = mHasRequirements;
@@ -1083,9 +1002,6 @@ void LLGLManager::initExtensions()
 	mHasCubeMap = FALSE;
 	mHasOcclusionQuery = FALSE;
 	mHasPointParameters = FALSE;
-	mHasShaderObjects = FALSE;
-	mHasVertexShader = FALSE;
-	mHasFragmentShader = FALSE;
 	mHasTextureRectangle = FALSE;
 #else // LL_MESA_HEADLESS //important, gGLHExts.mSysExts is uninitialized until after glh_init_extensions is called
 	mHasMultitexture = glh_init_extensions("GL_ARB_multitexture");
@@ -1105,8 +1021,9 @@ void LLGLManager::initExtensions()
 	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);
-	//mHasDepthClamp = ExtensionExists("GL_ARB_depth_clamp", gGLHExts.mSysExts) || ExtensionExists("GL_NV_depth_clamp", gGLHExts.mSysExts);
-	mHasDepthClamp = FALSE;
+    // NOTE: Using extensions breaks reflections when Shadows are set to projector.  See: SL-16727
+    //mHasDepthClamp = ExtensionExists("GL_ARB_depth_clamp", gGLHExts.mSysExts) || ExtensionExists("GL_NV_depth_clamp", gGLHExts.mSysExts);
+    mHasDepthClamp = FALSE;
 	// mask out FBO support when packed_depth_stencil isn't there 'cause we need it for LLRenderTarget -Brad
 #ifdef GL_ARB_framebuffer_object
 	mHasFramebufferObject = ExtensionExists("GL_ARB_framebuffer_object", gGLHExts.mSysExts);
@@ -1141,12 +1058,8 @@ void LLGLManager::initExtensions()
 	mHasDebugOutput = ExtensionExists("GL_ARB_debug_output", gGLHExts.mSysExts);
 	mHasTransformFeedback = mGLVersion >= 4.f ? TRUE : FALSE;
 #if !LL_DARWIN
-	mHasPointParameters = !mIsATI && ExtensionExists("GL_ARB_point_parameters", gGLHExts.mSysExts);
+	mHasPointParameters = ExtensionExists("GL_ARB_point_parameters", gGLHExts.mSysExts);
 #endif
-	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)
-		&& (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
@@ -1169,9 +1082,6 @@ void LLGLManager::initExtensions()
 		mHasCubeMap = FALSE;
 		mHasOcclusionQuery = FALSE;
 		mHasPointParameters = FALSE;
-		mHasShaderObjects = FALSE;
-		mHasVertexShader = FALSE;
-		mHasFragmentShader = FALSE;
 		LL_WARNS("RenderInit") << "GL extension support DISABLED via LL_GL_NOEXT" << LL_ENDL;
 	}
 	else if (getenv("LL_GL_BASICEXT"))	/* Flawfinder: ignore */
@@ -1184,9 +1094,6 @@ void LLGLManager::initExtensions()
 		mHasAnisotropic = FALSE;
 		//mHasCubeMap = FALSE; // apparently fatal on Intel 915 & similar
 		//mHasOcclusionQuery = FALSE; // source of many ATI system hangs
-		mHasShaderObjects = FALSE;
-		mHasVertexShader = FALSE;
-		mHasFragmentShader = FALSE;
 		mHasBlendFuncSeparate = FALSE;
 		LL_WARNS("RenderInit") << "GL extension support forced to SIMPLE level via LL_GL_BASICEXT" << LL_ENDL;
 	}
@@ -1208,9 +1115,6 @@ void LLGLManager::initExtensions()
 		if (strchr(blacklist,'j')) mHasCubeMap = FALSE;//S
 // 		if (strchr(blacklist,'k')) mHasATIVAO = FALSE;//S
 		if (strchr(blacklist,'l')) mHasOcclusionQuery = FALSE;
-		if (strchr(blacklist,'m')) mHasShaderObjects = FALSE;//S
-		if (strchr(blacklist,'n')) mHasVertexShader = FALSE;//S
-		if (strchr(blacklist,'o')) mHasFragmentShader = FALSE;//S
 		if (strchr(blacklist,'p')) mHasPointParameters = FALSE;//S
 		if (strchr(blacklist,'q')) mHasFramebufferObject = FALSE;//S
 		if (strchr(blacklist,'r')) mHasDrawBuffers = FALSE;//S
@@ -1257,18 +1161,6 @@ void LLGLManager::initExtensions()
 	{
 		LL_INFOS("RenderInit") << "Couldn't initialize GL_ARB_point_parameters" << LL_ENDL;
 	}
-	if (!mHasShaderObjects)
-	{
-		LL_INFOS("RenderInit") << "Couldn't initialize GL_ARB_shader_objects" << LL_ENDL;
-	}
-	if (!mHasVertexShader)
-	{
-		LL_INFOS("RenderInit") << "Couldn't initialize GL_ARB_vertex_shader" << LL_ENDL;
-	}
-	if (!mHasFragmentShader)
-	{
-		LL_INFOS("RenderInit") << "Couldn't initialize GL_ARB_fragment_shader" << LL_ENDL;
-	}
 	if (!mHasBlendFuncSeparate)
 	{
 		LL_INFOS("RenderInit") << "Couldn't initialize GL_EXT_blend_func_separate" << LL_ENDL;
@@ -1284,14 +1176,7 @@ void LLGLManager::initExtensions()
 		LL_INFOS("RenderInit") << "Disabling mip-map generation for Intel GPUs" << LL_ENDL;
 		mHasMipMapGeneration = FALSE;
 	}
-#if !LL_DARWIN
-	if (mIsATI && mHasMipMapGeneration)
-	{
-		LL_INFOS("RenderInit") << "Disabling mip-map generation for ATI GPUs (performance opt)" << LL_ENDL;
-		mHasMipMapGeneration = FALSE;
-	}
-#endif
-	
+
 	// Misc
 	glGetIntegerv(GL_MAX_ELEMENTS_VERTICES, (GLint*) &mGLMaxVertexRange);
 	glGetIntegerv(GL_MAX_ELEMENTS_INDICES, (GLint*) &mGLMaxIndexRange);
@@ -1436,134 +1321,132 @@ void LLGLManager::initExtensions()
 		glPointParameterfARB = (PFNGLPOINTPARAMETERFARBPROC)GLH_EXT_GET_PROC_ADDRESS("glPointParameterfARB");
 		glPointParameterfvARB = (PFNGLPOINTPARAMETERFVARBPROC)GLH_EXT_GET_PROC_ADDRESS("glPointParameterfvARB");
 	}
-	if (mHasShaderObjects)
-	{
-		glDeleteObjectARB = (PFNGLDELETEOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDeleteObjectARB");
-		glGetHandleARB = (PFNGLGETHANDLEARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetHandleARB");
-		glDetachObjectARB = (PFNGLDETACHOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDetachObjectARB");
-		glCreateShaderObjectARB = (PFNGLCREATESHADEROBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glCreateShaderObjectARB");
-		glShaderSourceARB = (PFNGLSHADERSOURCEARBPROC) GLH_EXT_GET_PROC_ADDRESS("glShaderSourceARB");
-		glCompileShaderARB = (PFNGLCOMPILESHADERARBPROC) GLH_EXT_GET_PROC_ADDRESS("glCompileShaderARB");
-		glCreateProgramObjectARB = (PFNGLCREATEPROGRAMOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glCreateProgramObjectARB");
-		glAttachObjectARB = (PFNGLATTACHOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glAttachObjectARB");
-		glLinkProgramARB = (PFNGLLINKPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glLinkProgramARB");
-		glUseProgramObjectARB = (PFNGLUSEPROGRAMOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUseProgramObjectARB");
-		glValidateProgramARB = (PFNGLVALIDATEPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glValidateProgramARB");
-		glUniform1fARB = (PFNGLUNIFORM1FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1fARB");
-		glUniform2fARB = (PFNGLUNIFORM2FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2fARB");
-		glUniform3fARB = (PFNGLUNIFORM3FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3fARB");
-		glUniform4fARB = (PFNGLUNIFORM4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4fARB");
-		glUniform1iARB = (PFNGLUNIFORM1IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1iARB");
-		glUniform2iARB = (PFNGLUNIFORM2IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2iARB");
-		glUniform3iARB = (PFNGLUNIFORM3IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3iARB");
-		glUniform4iARB = (PFNGLUNIFORM4IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4iARB");
-		glUniform1fvARB = (PFNGLUNIFORM1FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1fvARB");
-		glUniform2fvARB = (PFNGLUNIFORM2FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2fvARB");
-		glUniform3fvARB = (PFNGLUNIFORM3FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3fvARB");
-		glUniform4fvARB = (PFNGLUNIFORM4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4fvARB");
-		glUniform1ivARB = (PFNGLUNIFORM1IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1ivARB");
-		glUniform2ivARB = (PFNGLUNIFORM2IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2ivARB");
-		glUniform3ivARB = (PFNGLUNIFORM3IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3ivARB");
-		glUniform4ivARB = (PFNGLUNIFORM4IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4ivARB");
-		glUniformMatrix2fvARB = (PFNGLUNIFORMMATRIX2FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix2fvARB");
-		glUniformMatrix3fvARB = (PFNGLUNIFORMMATRIX3FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix3fvARB");
-		glUniformMatrix3x4fv = (PFNGLUNIFORMMATRIX3X4FVPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix3x4fv");
-		glUniformMatrix4fvARB = (PFNGLUNIFORMMATRIX4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix4fvARB");
-		glGetObjectParameterfvARB = (PFNGLGETOBJECTPARAMETERFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetObjectParameterfvARB");
-		glGetObjectParameterivARB = (PFNGLGETOBJECTPARAMETERIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetObjectParameterivARB");
-		glGetInfoLogARB = (PFNGLGETINFOLOGARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetInfoLogARB");
-		glGetAttachedObjectsARB = (PFNGLGETATTACHEDOBJECTSARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetAttachedObjectsARB");
-		glGetUniformLocationARB = (PFNGLGETUNIFORMLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetUniformLocationARB");
-		glGetActiveUniformARB = (PFNGLGETACTIVEUNIFORMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetActiveUniformARB");
-		glGetUniformfvARB = (PFNGLGETUNIFORMFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetUniformfvARB");
-		glGetUniformivARB = (PFNGLGETUNIFORMIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetUniformivARB");
-		glGetShaderSourceARB = (PFNGLGETSHADERSOURCEARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetShaderSourceARB");
-	}
-	if (mHasVertexShader)
-	{
-		LL_INFOS() << "initExtensions() VertexShader-related procs..." << LL_ENDL;
-
-        // nSight doesn't support use of ARB funcs that have been normalized in the API
-        if (!LLRender::sNsightDebugSupport)
-        {
-		glGetAttribLocationARB = (PFNGLGETATTRIBLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetAttribLocationARB");
-		glBindAttribLocationARB = (PFNGLBINDATTRIBLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glBindAttribLocationARB");
-        }
-        else
-        {
-            glGetAttribLocationARB = (PFNGLGETATTRIBLOCATIONARBPROC)GLH_EXT_GET_PROC_ADDRESS("glGetAttribLocation");
-            glBindAttribLocationARB = (PFNGLBINDATTRIBLOCATIONARBPROC)GLH_EXT_GET_PROC_ADDRESS("glBindAttribLocation");
-        }
 
-		glGetActiveAttribARB = (PFNGLGETACTIVEATTRIBARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetActiveAttribARB");
-		glVertexAttrib1dARB = (PFNGLVERTEXATTRIB1DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1dARB");
-		glVertexAttrib1dvARB = (PFNGLVERTEXATTRIB1DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1dvARB");
-		glVertexAttrib1fARB = (PFNGLVERTEXATTRIB1FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1fARB");
-		glVertexAttrib1fvARB = (PFNGLVERTEXATTRIB1FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1fvARB");
-		glVertexAttrib1sARB = (PFNGLVERTEXATTRIB1SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1sARB");
-		glVertexAttrib1svARB = (PFNGLVERTEXATTRIB1SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1svARB");
-		glVertexAttrib2dARB = (PFNGLVERTEXATTRIB2DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2dARB");
-		glVertexAttrib2dvARB = (PFNGLVERTEXATTRIB2DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2dvARB");
-		glVertexAttrib2fARB = (PFNGLVERTEXATTRIB2FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2fARB");
-		glVertexAttrib2fvARB = (PFNGLVERTEXATTRIB2FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2fvARB");
-		glVertexAttrib2sARB = (PFNGLVERTEXATTRIB2SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2sARB");
-		glVertexAttrib2svARB = (PFNGLVERTEXATTRIB2SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2svARB");
-		glVertexAttrib3dARB = (PFNGLVERTEXATTRIB3DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3dARB");
-		glVertexAttrib3dvARB = (PFNGLVERTEXATTRIB3DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3dvARB");
-		glVertexAttrib3fARB = (PFNGLVERTEXATTRIB3FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3fARB");
-		glVertexAttrib3fvARB = (PFNGLVERTEXATTRIB3FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3fvARB");
-		glVertexAttrib3sARB = (PFNGLVERTEXATTRIB3SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3sARB");
-		glVertexAttrib3svARB = (PFNGLVERTEXATTRIB3SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3svARB");
-		glVertexAttrib4nbvARB = (PFNGLVERTEXATTRIB4NBVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nbvARB");
-		glVertexAttrib4nivARB = (PFNGLVERTEXATTRIB4NIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nivARB");
-		glVertexAttrib4nsvARB = (PFNGLVERTEXATTRIB4NSVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nsvARB");
-		glVertexAttrib4nubARB = (PFNGLVERTEXATTRIB4NUBARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nubARB");
-		glVertexAttrib4nubvARB = (PFNGLVERTEXATTRIB4NUBVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nubvARB");
-		glVertexAttrib4nuivARB = (PFNGLVERTEXATTRIB4NUIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nuivARB");
-		glVertexAttrib4nusvARB = (PFNGLVERTEXATTRIB4NUSVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nusvARB");
-		glVertexAttrib4bvARB = (PFNGLVERTEXATTRIB4BVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4bvARB");
-		glVertexAttrib4dARB = (PFNGLVERTEXATTRIB4DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4dARB");
-		glVertexAttrib4dvARB = (PFNGLVERTEXATTRIB4DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4dvARB");
-		glVertexAttrib4fARB = (PFNGLVERTEXATTRIB4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4fARB");
-		glVertexAttrib4fvARB = (PFNGLVERTEXATTRIB4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4fvARB");
-		glVertexAttrib4ivARB = (PFNGLVERTEXATTRIB4IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4ivARB");
-		glVertexAttrib4sARB = (PFNGLVERTEXATTRIB4SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4sARB");
-		glVertexAttrib4svARB = (PFNGLVERTEXATTRIB4SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4svARB");
-		glVertexAttrib4ubvARB = (PFNGLVERTEXATTRIB4UBVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4ubvARB");
-		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");
-		glBindProgramARB = (PFNGLBINDPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glBindProgramARB");
-		glDeleteProgramsARB = (PFNGLDELETEPROGRAMSARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDeleteProgramsARB");
-		glGenProgramsARB = (PFNGLGENPROGRAMSARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGenProgramsARB");
-		glProgramEnvParameter4dARB = (PFNGLPROGRAMENVPARAMETER4DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4dARB");
-		glProgramEnvParameter4dvARB = (PFNGLPROGRAMENVPARAMETER4DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4dvARB");
-		glProgramEnvParameter4fARB = (PFNGLPROGRAMENVPARAMETER4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4fARB");
-		glProgramEnvParameter4fvARB = (PFNGLPROGRAMENVPARAMETER4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4fvARB");
-		glProgramLocalParameter4dARB = (PFNGLPROGRAMLOCALPARAMETER4DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4dARB");
-		glProgramLocalParameter4dvARB = (PFNGLPROGRAMLOCALPARAMETER4DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4dvARB");
-		glProgramLocalParameter4fARB = (PFNGLPROGRAMLOCALPARAMETER4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4fARB");
-		glProgramLocalParameter4fvARB = (PFNGLPROGRAMLOCALPARAMETER4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4fvARB");
-		glGetProgramEnvParameterdvARB = (PFNGLGETPROGRAMENVPARAMETERDVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramEnvParameterdvARB");
-		glGetProgramEnvParameterfvARB = (PFNGLGETPROGRAMENVPARAMETERFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramEnvParameterfvARB");
-		glGetProgramLocalParameterdvARB = (PFNGLGETPROGRAMLOCALPARAMETERDVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramLocalParameterdvARB");
-		glGetProgramLocalParameterfvARB = (PFNGLGETPROGRAMLOCALPARAMETERFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramLocalParameterfvARB");
-		glGetProgramivARB = (PFNGLGETPROGRAMIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramivARB");
-		glGetProgramStringARB = (PFNGLGETPROGRAMSTRINGARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramStringARB");
-		glGetVertexAttribdvARB = (PFNGLGETVERTEXATTRIBDVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetVertexAttribdvARB");
-		glGetVertexAttribfvARB = (PFNGLGETVERTEXATTRIBFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetVertexAttribfvARB");
-		glGetVertexAttribivARB = (PFNGLGETVERTEXATTRIBIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetVertexAttribivARB");
-		glGetVertexAttribPointervARB = (PFNGLGETVERTEXATTRIBPOINTERVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glgetVertexAttribPointervARB");
-		glIsProgramARB = (PFNGLISPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glIsProgramARB");
-	}
-	LL_DEBUGS("RenderInit") << "GL Probe: Got symbols" << LL_ENDL;
+    // Assume shader capabilities
+    glDeleteObjectARB         = (PFNGLDELETEOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDeleteObjectARB");
+    glGetHandleARB            = (PFNGLGETHANDLEARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetHandleARB");
+    glDetachObjectARB         = (PFNGLDETACHOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDetachObjectARB");
+    glCreateShaderObjectARB   = (PFNGLCREATESHADEROBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glCreateShaderObjectARB");
+    glShaderSourceARB         = (PFNGLSHADERSOURCEARBPROC) GLH_EXT_GET_PROC_ADDRESS("glShaderSourceARB");
+    glCompileShaderARB        = (PFNGLCOMPILESHADERARBPROC) GLH_EXT_GET_PROC_ADDRESS("glCompileShaderARB");
+    glCreateProgramObjectARB  = (PFNGLCREATEPROGRAMOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glCreateProgramObjectARB");
+    glAttachObjectARB         = (PFNGLATTACHOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glAttachObjectARB");
+    glLinkProgramARB          = (PFNGLLINKPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glLinkProgramARB");
+    glUseProgramObjectARB     = (PFNGLUSEPROGRAMOBJECTARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUseProgramObjectARB");
+    glValidateProgramARB      = (PFNGLVALIDATEPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glValidateProgramARB");
+    glUniform1fARB            = (PFNGLUNIFORM1FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1fARB");
+    glUniform2fARB            = (PFNGLUNIFORM2FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2fARB");
+    glUniform3fARB            = (PFNGLUNIFORM3FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3fARB");
+    glUniform4fARB            = (PFNGLUNIFORM4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4fARB");
+    glUniform1iARB            = (PFNGLUNIFORM1IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1iARB");
+    glUniform2iARB            = (PFNGLUNIFORM2IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2iARB");
+    glUniform3iARB            = (PFNGLUNIFORM3IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3iARB");
+    glUniform4iARB            = (PFNGLUNIFORM4IARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4iARB");
+    glUniform1fvARB           = (PFNGLUNIFORM1FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1fvARB");
+    glUniform2fvARB           = (PFNGLUNIFORM2FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2fvARB");
+    glUniform3fvARB           = (PFNGLUNIFORM3FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3fvARB");
+    glUniform4fvARB           = (PFNGLUNIFORM4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4fvARB");
+    glUniform1ivARB           = (PFNGLUNIFORM1IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform1ivARB");
+    glUniform2ivARB           = (PFNGLUNIFORM2IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform2ivARB");
+    glUniform3ivARB           = (PFNGLUNIFORM3IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform3ivARB");
+    glUniform4ivARB           = (PFNGLUNIFORM4IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniform4ivARB");
+    glUniformMatrix2fvARB     = (PFNGLUNIFORMMATRIX2FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix2fvARB");
+    glUniformMatrix3fvARB     = (PFNGLUNIFORMMATRIX3FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix3fvARB");
+    glUniformMatrix3x4fv      = (PFNGLUNIFORMMATRIX3X4FVPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix3x4fv");
+    glUniformMatrix4fvARB     = (PFNGLUNIFORMMATRIX4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glUniformMatrix4fvARB");
+    glGetObjectParameterfvARB = (PFNGLGETOBJECTPARAMETERFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetObjectParameterfvARB");
+    glGetObjectParameterivARB = (PFNGLGETOBJECTPARAMETERIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetObjectParameterivARB");
+    glGetInfoLogARB           = (PFNGLGETINFOLOGARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetInfoLogARB");
+    glGetAttachedObjectsARB   = (PFNGLGETATTACHEDOBJECTSARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetAttachedObjectsARB");
+    glGetUniformLocationARB   = (PFNGLGETUNIFORMLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetUniformLocationARB");
+    glGetActiveUniformARB     = (PFNGLGETACTIVEUNIFORMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetActiveUniformARB");
+    glGetUniformfvARB         = (PFNGLGETUNIFORMFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetUniformfvARB");
+    glGetUniformivARB         = (PFNGLGETUNIFORMIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetUniformivARB");
+    glGetShaderSourceARB      = (PFNGLGETSHADERSOURCEARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetShaderSourceARB");
+
+    LL_INFOS() << "initExtensions() VertexShader-related procs..." << LL_ENDL;
+
+    // nSight doesn't support use of ARB funcs that have been normalized in the API
+    if (!LLRender::sNsightDebugSupport)
+    {
+        glGetAttribLocationARB  = (PFNGLGETATTRIBLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetAttribLocationARB");
+        glBindAttribLocationARB = (PFNGLBINDATTRIBLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glBindAttribLocationARB");
+    }
+    else
+    {
+        glGetAttribLocationARB  = (PFNGLGETATTRIBLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetAttribLocation");
+        glBindAttribLocationARB = (PFNGLBINDATTRIBLOCATIONARBPROC) GLH_EXT_GET_PROC_ADDRESS("glBindAttribLocation");
+    }
+
+    glGetActiveAttribARB            = (PFNGLGETACTIVEATTRIBARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetActiveAttribARB");
+    glVertexAttrib1dARB             = (PFNGLVERTEXATTRIB1DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1dARB");
+    glVertexAttrib1dvARB            = (PFNGLVERTEXATTRIB1DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1dvARB");
+    glVertexAttrib1fARB             = (PFNGLVERTEXATTRIB1FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1fARB");
+    glVertexAttrib1fvARB            = (PFNGLVERTEXATTRIB1FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1fvARB");
+    glVertexAttrib1sARB             = (PFNGLVERTEXATTRIB1SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1sARB");
+    glVertexAttrib1svARB            = (PFNGLVERTEXATTRIB1SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib1svARB");
+    glVertexAttrib2dARB             = (PFNGLVERTEXATTRIB2DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2dARB");
+    glVertexAttrib2dvARB            = (PFNGLVERTEXATTRIB2DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2dvARB");
+    glVertexAttrib2fARB             = (PFNGLVERTEXATTRIB2FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2fARB");
+    glVertexAttrib2fvARB            = (PFNGLVERTEXATTRIB2FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2fvARB");
+    glVertexAttrib2sARB             = (PFNGLVERTEXATTRIB2SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2sARB");
+    glVertexAttrib2svARB            = (PFNGLVERTEXATTRIB2SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib2svARB");
+    glVertexAttrib3dARB             = (PFNGLVERTEXATTRIB3DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3dARB");
+    glVertexAttrib3dvARB            = (PFNGLVERTEXATTRIB3DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3dvARB");
+    glVertexAttrib3fARB             = (PFNGLVERTEXATTRIB3FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3fARB");
+    glVertexAttrib3fvARB            = (PFNGLVERTEXATTRIB3FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3fvARB");
+    glVertexAttrib3sARB             = (PFNGLVERTEXATTRIB3SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3sARB");
+    glVertexAttrib3svARB            = (PFNGLVERTEXATTRIB3SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib3svARB");
+    glVertexAttrib4nbvARB           = (PFNGLVERTEXATTRIB4NBVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nbvARB");
+    glVertexAttrib4nivARB           = (PFNGLVERTEXATTRIB4NIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nivARB");
+    glVertexAttrib4nsvARB           = (PFNGLVERTEXATTRIB4NSVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nsvARB");
+    glVertexAttrib4nubARB           = (PFNGLVERTEXATTRIB4NUBARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nubARB");
+    glVertexAttrib4nubvARB          = (PFNGLVERTEXATTRIB4NUBVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nubvARB");
+    glVertexAttrib4nuivARB          = (PFNGLVERTEXATTRIB4NUIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nuivARB");
+    glVertexAttrib4nusvARB          = (PFNGLVERTEXATTRIB4NUSVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4nusvARB");
+    glVertexAttrib4bvARB            = (PFNGLVERTEXATTRIB4BVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4bvARB");
+    glVertexAttrib4dARB             = (PFNGLVERTEXATTRIB4DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4dARB");
+    glVertexAttrib4dvARB            = (PFNGLVERTEXATTRIB4DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4dvARB");
+    glVertexAttrib4fARB             = (PFNGLVERTEXATTRIB4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4fARB");
+    glVertexAttrib4fvARB            = (PFNGLVERTEXATTRIB4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4fvARB");
+    glVertexAttrib4ivARB            = (PFNGLVERTEXATTRIB4IVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4ivARB");
+    glVertexAttrib4sARB             = (PFNGLVERTEXATTRIB4SARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4sARB");
+    glVertexAttrib4svARB            = (PFNGLVERTEXATTRIB4SVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4svARB");
+    glVertexAttrib4ubvARB           = (PFNGLVERTEXATTRIB4UBVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glVertexAttrib4ubvARB");
+    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");
+    glBindProgramARB                = (PFNGLBINDPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glBindProgramARB");
+    glDeleteProgramsARB             = (PFNGLDELETEPROGRAMSARBPROC) GLH_EXT_GET_PROC_ADDRESS("glDeleteProgramsARB");
+    glGenProgramsARB                = (PFNGLGENPROGRAMSARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGenProgramsARB");
+    glProgramEnvParameter4dARB      = (PFNGLPROGRAMENVPARAMETER4DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4dARB");
+    glProgramEnvParameter4dvARB     = (PFNGLPROGRAMENVPARAMETER4DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4dvARB");
+    glProgramEnvParameter4fARB      = (PFNGLPROGRAMENVPARAMETER4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4fARB");
+    glProgramEnvParameter4fvARB     = (PFNGLPROGRAMENVPARAMETER4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramEnvParameter4fvARB");
+    glProgramLocalParameter4dARB    = (PFNGLPROGRAMLOCALPARAMETER4DARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4dARB");
+    glProgramLocalParameter4dvARB   = (PFNGLPROGRAMLOCALPARAMETER4DVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4dvARB");
+    glProgramLocalParameter4fARB    = (PFNGLPROGRAMLOCALPARAMETER4FARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4fARB");
+    glProgramLocalParameter4fvARB   = (PFNGLPROGRAMLOCALPARAMETER4FVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glProgramLocalParameter4fvARB");
+    glGetProgramEnvParameterdvARB   = (PFNGLGETPROGRAMENVPARAMETERDVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramEnvParameterdvARB");
+    glGetProgramEnvParameterfvARB   = (PFNGLGETPROGRAMENVPARAMETERFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramEnvParameterfvARB");
+    glGetProgramLocalParameterdvARB = (PFNGLGETPROGRAMLOCALPARAMETERDVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramLocalParameterdvARB");
+    glGetProgramLocalParameterfvARB = (PFNGLGETPROGRAMLOCALPARAMETERFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramLocalParameterfvARB");
+    glGetProgramivARB               = (PFNGLGETPROGRAMIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramivARB");
+    glGetProgramStringARB           = (PFNGLGETPROGRAMSTRINGARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetProgramStringARB");
+    glGetVertexAttribdvARB          = (PFNGLGETVERTEXATTRIBDVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetVertexAttribdvARB");
+    glGetVertexAttribfvARB          = (PFNGLGETVERTEXATTRIBFVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetVertexAttribfvARB");
+    glGetVertexAttribivARB          = (PFNGLGETVERTEXATTRIBIVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glGetVertexAttribivARB");
+    glGetVertexAttribPointervARB    = (PFNGLGETVERTEXATTRIBPOINTERVARBPROC) GLH_EXT_GET_PROC_ADDRESS("glgetVertexAttribPointervARB");
+    glIsProgramARB                  = (PFNGLISPROGRAMARBPROC) GLH_EXT_GET_PROC_ADDRESS("glIsProgramARB");
+
+    LL_DEBUGS("RenderInit") << "GL Probe: Got symbols" << LL_ENDL;
 #endif
 
-	mInited = TRUE;
+    mInited = TRUE;
 }
 
 void rotate_quat(LLQuaternion& rotation)
@@ -1974,206 +1857,30 @@ void LLGLState::checkTextureChannels(const std::string& msg)
 #endif
 }
 
-void LLGLState::checkClientArrays(const std::string& msg, U32 data_mask)
-{
-	if (!gDebugGL || LLGLSLShader::sNoFixedFunction)
-	{
-		return;
-	}
-
-	stop_glerror();
-	BOOL error = FALSE;
-
-	GLint active_texture;
-	glGetIntegerv(GL_CLIENT_ACTIVE_TEXTURE_ARB, &active_texture);
-
-	if (active_texture != GL_TEXTURE0_ARB)
-	{
-		LL_WARNS() << "Client active texture corrupted: " << active_texture << LL_ENDL;
-		if (gDebugSession)
-		{
-			gFailLog << "Client active texture corrupted: " << active_texture << std::endl;
-		}
-		error = TRUE;
-	}
-
-	/*glGetIntegerv(GL_ACTIVE_TEXTURE_ARB, &active_texture);
-	if (active_texture != GL_TEXTURE0_ARB)
-	{
-		LL_WARNS() << "Active texture corrupted: " << active_texture << LL_ENDL;
-		if (gDebugSession)
-		{
-			gFailLog << "Active texture corrupted: " << active_texture << std::endl;
-		}
-		error = TRUE;
-	}*/
-
-	static const char* label[] =
-	{
-		"GL_VERTEX_ARRAY",
-		"GL_NORMAL_ARRAY",
-		"GL_COLOR_ARRAY",
-		"GL_TEXTURE_COORD_ARRAY"
-	};
-
-	static GLint value[] =
-	{
-		GL_VERTEX_ARRAY,
-		GL_NORMAL_ARRAY,
-		GL_COLOR_ARRAY,
-		GL_TEXTURE_COORD_ARRAY
-	};
-
-	static const U32 mask[] = 
-	{ //copied from llvertexbuffer.h
-		0x0001, //MAP_VERTEX,
-		0x0002, //MAP_NORMAL,
-		0x0010, //MAP_COLOR,
-		0x0004, //MAP_TEXCOORD
-	};
-
-
-	for (S32 j = 1; j < 4; j++)
-	{
-		if (glIsEnabled(value[j]))
-		{
-			if (!(mask[j] & data_mask))
-			{
-				error = TRUE;
-				LL_WARNS("RenderState") << "GL still has " << label[j] << " enabled." << LL_ENDL;
-				if (gDebugSession)
-				{
-					gFailLog << "GL still has " << label[j] << " enabled." << std::endl;
-				}
-			}
-		}
-		else
-		{
-			if (mask[j] & data_mask)
-			{
-				error = TRUE;
-				LL_WARNS("RenderState") << "GL does not have " << label[j] << " enabled." << LL_ENDL;
-				if (gDebugSession)
-				{
-					gFailLog << "GL does not have " << label[j] << " enabled." << std::endl;
-				}
-			}
-		}
-	}
-
-	glClientActiveTextureARB(GL_TEXTURE1_ARB);
-	gGL.getTexUnit(1)->activate();
-	if (glIsEnabled(GL_TEXTURE_COORD_ARRAY))
-	{
-		if (!(data_mask & 0x0008))
-		{
-			error = TRUE;
-			LL_WARNS("RenderState") << "GL still has GL_TEXTURE_COORD_ARRAY enabled on channel 1." << LL_ENDL;
-			if (gDebugSession)
-			{
-				gFailLog << "GL still has GL_TEXTURE_COORD_ARRAY enabled on channel 1." << std::endl;
-			}
-		}
-	}
-	else
-	{
-		if (data_mask & 0x0008)
-		{
-			error = TRUE;
-			LL_WARNS("RenderState") << "GL does not have GL_TEXTURE_COORD_ARRAY enabled on channel 1." << LL_ENDL;
-			if (gDebugSession)
-			{
-				gFailLog << "GL does not have GL_TEXTURE_COORD_ARRAY enabled on channel 1." << std::endl;
-			}
-		}
-	}
-
-	/*if (glIsEnabled(GL_TEXTURE_2D))
-	{
-		if (!(data_mask & 0x0008))
-		{
-			error = TRUE;
-			LL_WARNS("RenderState") << "GL still has GL_TEXTURE_2D enabled on channel 1." << LL_ENDL;
-			if (gDebugSession)
-			{
-				gFailLog << "GL still has GL_TEXTURE_2D enabled on channel 1." << std::endl;
-			}
-		}
-	}
-	else
-	{
-		if (data_mask & 0x0008)
-		{
-			error = TRUE;
-			LL_WARNS("RenderState") << "GL does not have GL_TEXTURE_2D enabled on channel 1." << LL_ENDL;
-			if (gDebugSession)
-			{
-				gFailLog << "GL does not have GL_TEXTURE_2D enabled on channel 1." << std::endl;
-			}
-		}
-	}*/
-
-	glClientActiveTextureARB(GL_TEXTURE0_ARB);
-	gGL.getTexUnit(0)->activate();
-
-	if (gGLManager.mHasVertexShader && LLGLSLShader::sNoFixedFunction)
-	{	//make sure vertex attribs are all disabled
-		GLint count;
-		glGetIntegerv(GL_MAX_VERTEX_ATTRIBS_ARB, &count);
-		for (GLint i = 0; i < count; i++)
-		{
-			GLint enabled;
-			glGetVertexAttribivARB((GLuint) i, GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB, &enabled);
-			if (enabled)
-			{
-				error = TRUE;
-				LL_WARNS("RenderState") << "GL still has vertex attrib array " << i << " enabled." << LL_ENDL;
-				if (gDebugSession)
-				{
-					gFailLog <<  "GL still has vertex attrib array " << i << " enabled." << std::endl;
-				}
-			}
-		}
-	}
-
-	if (error)
-	{
-		if (gDebugSession)
-		{
-			ll_fail("LLGLState::checkClientArrays failed.");
-		}
-		else
-		{
-			LL_GL_ERRS << "GL client array corruption detected.  " << msg << LL_ENDL;
-		}
-	}
-}
-
 ///////////////////////////////////////////////////////////////////////
 
 LLGLState::LLGLState(LLGLenum state, S32 enabled) :
 	mState(state), mWasEnabled(FALSE), mIsEnabled(FALSE)
 {
-	if (LLGLSLShader::sNoFixedFunction)
-	{ //always ignore state that's deprecated post GL 3.0
-		switch (state)
-		{
-			case GL_ALPHA_TEST:
-			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:
-			case GL_POLYGON_STIPPLE:
-				mState = 0;
-				break;
-		}
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_PIPELINE;
+	switch (state)
+	{
+		case GL_ALPHA_TEST:
+		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:
+		case GL_POLYGON_STIPPLE:
+			mState = 0;
+			break;
 	}
 
+
 	stop_glerror();
 	if (mState)
 	{
@@ -2212,6 +1919,7 @@ void LLGLState::setEnabled(S32 enabled)
 
 LLGLState::~LLGLState() 
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_PIPELINE;
 	stop_glerror();
 	if (mState)
 	{
@@ -2398,7 +2106,8 @@ LLGLUserClipPlane::LLGLUserClipPlane(const LLPlane& p, const glh::matrix4f& mode
 		mModelview = modelview;
 		mProjection = projection;
 
-		setPlane(p[0], p[1], p[2], p[3]);
+        //flip incoming LLPlane to get consistent behavior compared to frustum culling
+		setPlane(-p[0], -p[1], -p[2], -p[3]);
 	}
 }
 
@@ -2444,94 +2153,6 @@ LLGLUserClipPlane::~LLGLUserClipPlane()
 	disable();
 }
 
-LLGLNamePool::LLGLNamePool()
-{
-}
-
-LLGLNamePool::~LLGLNamePool()
-{
-}
-
-void LLGLNamePool::upkeep()
-{
-	std::sort(mNameList.begin(), mNameList.end(), CompareUsed());
-}
-
-void LLGLNamePool::cleanup()
-{
-	for (name_list_t::iterator iter = mNameList.begin(); iter != mNameList.end(); ++iter)
-	{
-		releaseName(iter->name);
-	}
-
-	mNameList.clear();
-}
-
-GLuint LLGLNamePool::allocate()
-{
-#if LL_GL_NAME_POOLING
-	for (name_list_t::iterator iter = mNameList.begin(); iter != mNameList.end(); ++iter)
-	{
-		if (!iter->used)
-		{
-			iter->used = TRUE;
-			return iter->name;
-		}
-	}
-
-	NameEntry entry;
-	entry.name = allocateName();
-	entry.used = TRUE;
-	mNameList.push_back(entry);
-
-	return entry.name;
-#else
-	return allocateName();
-#endif
-}
-
-void LLGLNamePool::release(GLuint name)
-{
-#if LL_GL_NAME_POOLING
-	for (name_list_t::iterator iter = mNameList.begin(); iter != mNameList.end(); ++iter)
-	{
-		if (iter->name == name)
-		{
-			if (iter->used)
-			{
-				iter->used = FALSE;
-				return;
-			}
-			else
-			{
-				LL_ERRS() << "Attempted to release a pooled name that is not in use!" << LL_ENDL;
-			}
-		}
-	}
-	LL_ERRS() << "Attempted to release a non pooled name!" << LL_ENDL;
-#else
-	releaseName(name);
-#endif
-}
-
-//static
-void LLGLNamePool::upkeepPools()
-{
-	for (auto& pool : instance_snapshot())
-	{
-		pool.upkeep();
-	}
-}
-
-//static
-void LLGLNamePool::cleanupPools()
-{
-	for (auto& pool : instance_snapshot())
-	{
-		pool.cleanup();
-	}
-}
-
 LLGLDepthTest::LLGLDepthTest(GLboolean depth_enabled, GLboolean write_enabled, GLenum depth_func)
 : mPrevDepthEnabled(sDepthEnabled), mPrevDepthFunc(sDepthFunc), mPrevWriteEnabled(sWriteEnabled)
 {
@@ -2723,22 +2344,10 @@ LLGLSPipelineSkyBox::LLGLSPipelineSkyBox()
 , mCullFace(GL_CULL_FACE)
 , mSquashClip()
 { 
-    if (!LLGLSLShader::sNoFixedFunction)
-    {
-        glDisable(GL_LIGHTING);
-        glDisable(GL_FOG);
-        glDisable(GL_CLIP_PLANE0);
-    }
 }
 
 LLGLSPipelineSkyBox::~LLGLSPipelineSkyBox()
 {
-    if (!LLGLSLShader::sNoFixedFunction)
-    {
-        glEnable(GL_LIGHTING);
-        glEnable(GL_FOG);
-        glEnable(GL_CLIP_PLANE0);
-    }
 }
 
 LLGLSPipelineDepthTestSkyBox::LLGLSPipelineDepthTestSkyBox(bool depth_test, bool depth_write)
@@ -2765,3 +2374,4 @@ extern "C"
 }
 #endif
 
+
diff --git a/indra/llrender/llgl.h b/indra/llrender/llgl.h
index a07e2d9bb0..52338364e6 100644
--- a/indra/llrender/llgl.h
+++ b/indra/llrender/llgl.h
@@ -94,9 +94,6 @@ public:
 	BOOL mHasMapBufferRange;
 	BOOL mHasFlushBufferRange;
 	BOOL mHasPBuffer;
-	BOOL mHasShaderObjects;
-	BOOL mHasVertexShader;
-	BOOL mHasFragmentShader;
 	S32  mNumTextureImageUnits;
 	BOOL mHasOcclusionQuery;
 	BOOL mHasTimerQuery;
@@ -122,14 +119,9 @@ public:
     BOOL mHasTexturesRGBDecode;
 
 	// Vendor-specific extensions
-	BOOL mIsATI;
+	BOOL mIsAMD;
 	BOOL mIsNVIDIA;
 	BOOL mIsIntel;
-	BOOL mIsGF2or4MX;
-	BOOL mIsGF3;
-	BOOL mIsGFFX;
-	BOOL mATIOffsetVerticalLines;
-	BOOL mATIOldDriver;
 
 #if LL_DARWIN
 	// Needed to distinguish problem cards on older Macs that break with Materials
@@ -269,7 +261,6 @@ 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 = 0);
 	
 protected:
 	static boost::unordered_map<LLGLenum, LLGLboolean> sStateMap;
@@ -376,51 +367,6 @@ public:
 };
 
 /*
-	Generic pooling scheme for things which use GL names (used for occlusion queries and vertex buffer objects).
-	Prevents thrashing of GL name caches by avoiding calls to glGenFoo and glDeleteFoo.
-*/
-class LLGLNamePool : public LLInstanceTracker<LLGLNamePool>
-{
-public:
-	typedef LLInstanceTracker<LLGLNamePool> tracker_t;
-
-	struct NameEntry
-	{
-		GLuint name;
-		BOOL used;
-	};
-
-	struct CompareUsed
-	{
-		bool operator()(const NameEntry& lhs, const NameEntry& rhs)
-		{
-			return lhs.used < rhs.used;  //FALSE entries first
-		}
-	};
-
-	typedef std::vector<NameEntry> name_list_t;
-	name_list_t mNameList;
-
-	LLGLNamePool();
-	virtual ~LLGLNamePool();
-	
-	void upkeep();
-	void cleanup();
-	
-	GLuint allocate();
-	void release(GLuint name);
-	
-	static void upkeepPools();
-	static void cleanupPools();
-
-protected:
-	typedef std::vector<LLGLNamePool*> pool_list_t;
-	
-	virtual GLuint allocateName() = 0;
-	virtual void releaseName(GLuint name) = 0;
-};
-
-/*
 	Interface for objects that need periodic GL updates applied to them.
 	Used to synchronize GL updates with GL thread.
 */
@@ -488,6 +434,7 @@ void parse_gl_version( S32* major, S32* minor, S32* release, std::string* vendor
 
 extern BOOL gClothRipple;
 extern BOOL gHeadlessClient;
+extern BOOL gNonInteractive;
 extern BOOL gGLActive;
 
 // Deal with changing glext.h definitions for newer SDK versions, specifically
diff --git a/indra/llrender/llglheaders.h b/indra/llrender/llglheaders.h
index 6bca3623e0..3d93cc0762 100644
--- a/indra/llrender/llglheaders.h
+++ b/indra/llrender/llglheaders.h
@@ -812,4 +812,23 @@ extern void glGetBufferPointervARB (GLenum, GLenum, GLvoid* *);
 #define GL_RENDERBUFFER_FREE_MEMORY_ATI            0x87FD
 #endif
 
+#if defined(TRACY_ENABLE) && LL_PROFILER_ENABLE_TRACY_OPENGL
+    // Tracy uses the following:
+    //    glGenQueries
+    //    glGetQueryiv
+    //    glGetQueryObjectiv
+    #define glGenQueries        glGenQueriesARB
+    #define glGetQueryiv        glGetQueryivARB
+    #define glGetQueryObjectiv  glGetQueryObjectivARB
+    #include <tracy/TracyOpenGL.hpp>
+
+    #define LL_PROFILER_GPU_ZONEC(name,color) TracyGpuZoneC(name,color);
+    #define LL_PROFILER_GPU_COLLECT           TracyGpuCollect
+    #define LL_PROFILER_GPU_CONTEXT           TracyGpuContext
+#else
+    #define LL_PROFILER_GPU_ZONEC(name,color) (void)name;(void)color;
+    #define LL_PROFILER_GPU_COLLECT
+    #define LL_PROFILER_GPU_CONTEXT
+#endif
+
 #endif // LL_LLGLHEADERS_H
diff --git a/indra/llrender/llglslshader.cpp b/indra/llrender/llglslshader.cpp
index 42b8c35442..3a0745106f 100644
--- a/indra/llrender/llglslshader.cpp
+++ b/indra/llrender/llglslshader.cpp
@@ -50,7 +50,6 @@ using std::string;
 GLhandleARB LLGLSLShader::sCurBoundShader = 0;
 LLGLSLShader* LLGLSLShader::sCurBoundShaderPtr = NULL;
 S32 LLGLSLShader::sIndexedTextureChannels = 0;
-bool LLGLSLShader::sNoFixedFunction = false;
 bool LLGLSLShader::sProfileEnabled = false;
 std::set<LLGLSLShader*> LLGLSLShader::sInstances;
 U64 LLGLSLShader::sTotalTimeElapsed = 0;
@@ -223,6 +222,7 @@ void LLGLSLShader::dumpStats()
 //static
 void LLGLSLShader::startProfile()
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
     if (sProfileEnabled && sCurBoundShaderPtr)
     {
         sCurBoundShaderPtr->placeProfileQuery();
@@ -233,6 +233,7 @@ void LLGLSLShader::startProfile()
 //static
 void LLGLSLShader::stopProfile(U32 count, U32 mode)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
     if (sProfileEnabled && sCurBoundShaderPtr)
     {
         sCurBoundShaderPtr->readProfileQuery(count, mode);
@@ -399,6 +400,8 @@ BOOL LLGLSLShader::createShader(std::vector<LLStaticHashedString> * attributes,
                                 U32 varying_count,
                                 const char** varyings)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     unloadInternal();
 
     sInstances.insert(this);
@@ -603,6 +606,8 @@ void LLGLSLShader::attachObjects(GLhandleARB* objects, S32 count)
 
 BOOL LLGLSLShader::mapAttributes(const std::vector<LLStaticHashedString> * attributes)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     //before linking, make sure reserved attributes always have consistent locations
     for (U32 i = 0; i < LLShaderMgr::instance()->mReservedAttribs.size(); i++)
     {
@@ -664,6 +669,8 @@ BOOL LLGLSLShader::mapAttributes(const std::vector<LLStaticHashedString> * attri
 
 void LLGLSLShader::mapUniform(GLint index, const vector<LLStaticHashedString> * uniforms)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if (index == -1)
     {
         return;
@@ -790,6 +797,8 @@ void LLGLSLShader::removePermutation(std::string name)
 
 GLint LLGLSLShader::mapUniformTextureChannel(GLint location, GLenum type)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if ((type >= GL_SAMPLER_1D_ARB && type <= GL_SAMPLER_2D_RECT_SHADOW_ARB) ||
         type == GL_SAMPLER_2D_MULTISAMPLE)
     {   //this here is a texture
@@ -802,7 +811,9 @@ GLint LLGLSLShader::mapUniformTextureChannel(GLint location, GLenum type)
 
 BOOL LLGLSLShader::mapUniforms(const vector<LLStaticHashedString> * uniforms)
 {
-	BOOL res = TRUE;
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
+    BOOL res = TRUE;
 
 	mTotalUniformSize = 0;
 	mActiveTextureChannels = 0;
@@ -945,6 +956,8 @@ BOOL LLGLSLShader::mapUniforms(const vector<LLStaticHashedString> * uniforms)
 
 BOOL LLGLSLShader::link(BOOL suppress_errors)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     BOOL success = LLShaderMgr::instance()->linkProgramObject(mProgramObject, suppress_errors);
 
     if (!success && !suppress_errors)
@@ -957,56 +970,65 @@ BOOL LLGLSLShader::link(BOOL suppress_errors)
 
 void LLGLSLShader::bind()
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     gGL.flush();
-    if (gGLManager.mHasShaderObjects)
+
+    if (sCurBoundShader != mProgramObject)  // Don't re-bind current shader
     {
         LLVertexBuffer::unbind();
         glUseProgramObjectARB(mProgramObject);
         sCurBoundShader = mProgramObject;
         sCurBoundShaderPtr = this;
-        if (mUniformsDirty)
-        {
-            LLShaderMgr::instance()->updateShaderUniforms(this);
-            mUniformsDirty = FALSE;
-        }
+    }
+
+    if (mUniformsDirty)
+    {
+        LLShaderMgr::instance()->updateShaderUniforms(this);
+        mUniformsDirty = FALSE;
     }
 }
 
-void LLGLSLShader::unbind()
+void LLGLSLShader::bind(bool rigged)
 {
-    gGL.flush();
-    if (gGLManager.mHasShaderObjects)
+    if (rigged)
     {
-        stop_glerror();
-        if (gGLManager.mIsNVIDIA)
-        {
-            for (U32 i = 0; i < mAttribute.size(); ++i)
-            {
-                vertexAttrib4f(i, 0,0,0,1);
-                stop_glerror();
-            }
-        }
-        LLVertexBuffer::unbind();
-        glUseProgramObjectARB(0);
-        sCurBoundShader = 0;
-        sCurBoundShaderPtr = NULL;
-        stop_glerror();
+        llassert(mRiggedVariant);
+        mRiggedVariant->bind();
+    }
+    else
+    {
+        bind();
     }
 }
 
+void LLGLSLShader::unbind()
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
+    gGL.flush();
+    stop_glerror();
+    LLVertexBuffer::unbind();
+    glUseProgramObjectARB(0);
+    sCurBoundShader = 0;
+    sCurBoundShaderPtr = NULL;
+    stop_glerror();
+}
+
 void LLGLSLShader::bindNoShader(void)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     LLVertexBuffer::unbind();
-    if (gGLManager.mHasShaderObjects)
-    {
-        glUseProgramObjectARB(0);
-        sCurBoundShader = 0;
-        sCurBoundShaderPtr = NULL;
-    }
+    glUseProgramObjectARB(0);
+    sCurBoundShader = 0;
+    sCurBoundShaderPtr = NULL;
 }
 
 S32 LLGLSLShader::bindTexture(const std::string &uniform, LLTexture *texture, LLTexUnit::eTextureType mode, LLTexUnit::eTextureColorSpace colorspace)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     S32 channel = 0;
     channel = getUniformLocation(uniform);
     
@@ -1015,6 +1037,8 @@ S32 LLGLSLShader::bindTexture(const std::string &uniform, LLTexture *texture, LL
 
 S32 LLGLSLShader::bindTexture(S32 uniform, LLTexture *texture, LLTexUnit::eTextureType mode, LLTexUnit::eTextureColorSpace colorspace)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if (uniform < 0 || uniform >= (S32)mTexture.size())
     {
         LL_SHADER_UNIFORM_ERRS() << "Uniform out of range: " << uniform << LL_ENDL;
@@ -1025,7 +1049,7 @@ S32 LLGLSLShader::bindTexture(S32 uniform, LLTexture *texture, LLTexUnit::eTextu
     
     if (uniform > -1)
     {
-        gGL.getTexUnit(uniform)->bind(texture, mode);
+        gGL.getTexUnit(uniform)->bindFast(texture);
         gGL.getTexUnit(uniform)->setTextureColorSpace(colorspace);
     }
     
@@ -1034,6 +1058,8 @@ S32 LLGLSLShader::bindTexture(S32 uniform, LLTexture *texture, LLTexUnit::eTextu
 
 S32 LLGLSLShader::unbindTexture(const std::string &uniform, LLTexUnit::eTextureType mode)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     S32 channel = 0;
     channel = getUniformLocation(uniform);
     
@@ -1042,6 +1068,8 @@ S32 LLGLSLShader::unbindTexture(const std::string &uniform, LLTexUnit::eTextureT
 
 S32 LLGLSLShader::unbindTexture(S32 uniform, LLTexUnit::eTextureType mode)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if (uniform < 0 || uniform >= (S32)mTexture.size())
     {
         LL_SHADER_UNIFORM_ERRS() << "Uniform out of range: " << uniform << LL_ENDL;
@@ -1052,7 +1080,7 @@ S32 LLGLSLShader::unbindTexture(S32 uniform, LLTexUnit::eTextureType mode)
     
     if (uniform > -1)
     {
-        gGL.getTexUnit(uniform)->unbind(mode);
+        gGL.getTexUnit(uniform)->unbindFast(mode);
     }
     
     return uniform;
@@ -1060,6 +1088,8 @@ S32 LLGLSLShader::unbindTexture(S32 uniform, LLTexUnit::eTextureType mode)
 
 S32 LLGLSLShader::enableTexture(S32 uniform, LLTexUnit::eTextureType mode, LLTexUnit::eTextureColorSpace space)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if (uniform < 0 || uniform >= (S32)mTexture.size())
     {
         LL_SHADER_UNIFORM_ERRS() << "Uniform out of range: " << uniform << LL_ENDL;
@@ -1077,6 +1107,8 @@ S32 LLGLSLShader::enableTexture(S32 uniform, LLTexUnit::eTextureType mode, LLTex
 
 S32 LLGLSLShader::disableTexture(S32 uniform, LLTexUnit::eTextureType mode, LLTexUnit::eTextureColorSpace space)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if (uniform < 0 || uniform >= (S32)mTexture.size())
     {
         LL_SHADER_UNIFORM_ERRS() << "Uniform out of range: " << uniform << LL_ENDL;
@@ -1104,6 +1136,7 @@ S32 LLGLSLShader::disableTexture(S32 uniform, LLTexUnit::eTextureType mode, LLTe
 
 void LLGLSLShader::uniform1i(U32 index, GLint x)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER
     if (mProgramObject)
     {   
         if (mUniform.size() <= index)
@@ -1114,7 +1147,7 @@ void LLGLSLShader::uniform1i(U32 index, GLint x)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             if (iter == mValue.end() || iter->second.mV[0] != x)
             {
                 glUniform1iARB(mUniform[index], x);
@@ -1126,6 +1159,7 @@ void LLGLSLShader::uniform1i(U32 index, GLint x)
 
 void LLGLSLShader::uniform1f(U32 index, GLfloat x)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER
     if (mProgramObject)
     {   
         if (mUniform.size() <= index)
@@ -1136,7 +1170,7 @@ void LLGLSLShader::uniform1f(U32 index, GLfloat x)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             if (iter == mValue.end() || iter->second.mV[0] != x)
             {
                 glUniform1fARB(mUniform[index], x);
@@ -1158,7 +1192,7 @@ void LLGLSLShader::uniform2f(U32 index, GLfloat x, GLfloat y)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(x,y,0.f,0.f);
             if (iter == mValue.end() || shouldChange(iter->second,vec))
             {
@@ -1181,7 +1215,7 @@ void LLGLSLShader::uniform3f(U32 index, GLfloat x, GLfloat y, GLfloat z)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(x,y,z,0.f);
             if (iter == mValue.end() || shouldChange(iter->second,vec))
             {
@@ -1204,7 +1238,7 @@ void LLGLSLShader::uniform4f(U32 index, GLfloat x, GLfloat y, GLfloat z, GLfloat
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(x,y,z,w);
             if (iter == mValue.end() || shouldChange(iter->second,vec))
             {
@@ -1227,7 +1261,7 @@ void LLGLSLShader::uniform1iv(U32 index, U32 count, const GLint* v)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(v[0],0.f,0.f,0.f);
             if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
             {
@@ -1250,7 +1284,7 @@ void LLGLSLShader::uniform1fv(U32 index, U32 count, const GLfloat* v)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(v[0],0.f,0.f,0.f);
             if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
             {
@@ -1273,7 +1307,7 @@ void LLGLSLShader::uniform2fv(U32 index, U32 count, const GLfloat* v)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(v[0],v[1],0.f,0.f);
             if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
             {
@@ -1296,7 +1330,7 @@ void LLGLSLShader::uniform3fv(U32 index, U32 count, const GLfloat* v)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(v[0],v[1],v[2],0.f);
             if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
             {
@@ -1319,10 +1353,11 @@ void LLGLSLShader::uniform4fv(U32 index, U32 count, const GLfloat* v)
 
         if (mUniform[index] >= 0)
         {
-            std::map<GLint, LLVector4>::iterator iter = mValue.find(mUniform[index]);
+            const auto& iter = mValue.find(mUniform[index]);
             LLVector4 vec(v[0],v[1],v[2],v[3]);
             if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
             {
+                LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
                 glUniform4fvARB(mUniform[index], count, v);
                 mValue[mUniform[index]] = vec;
             }
@@ -1366,6 +1401,8 @@ void LLGLSLShader::uniformMatrix3fv(U32 index, U32 count, GLboolean transpose, c
 
 void LLGLSLShader::uniformMatrix3x4fv(U32 index, U32 count, GLboolean transpose, const GLfloat *v)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
 	if (mProgramObject)
 	{	
 		if (mUniform.size() <= index)
@@ -1400,6 +1437,8 @@ void LLGLSLShader::uniformMatrix4fv(U32 index, U32 count, GLboolean transpose, c
 
 GLint LLGLSLShader::getUniformLocation(const LLStaticHashedString& uniform)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     GLint ret = -1;
     if (mProgramObject)
     {
@@ -1424,10 +1463,16 @@ GLint LLGLSLShader::getUniformLocation(const LLStaticHashedString& uniform)
 
 GLint LLGLSLShader::getUniformLocation(U32 index)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     GLint ret = -1;
     if (mProgramObject)
     {
-        llassert(index < mUniform.size());
+        if (index >= mUniform.size())
+        {
+            LL_WARNS_ONCE("Shader") << "Uniform index " << index << " out of bounds " << (S32)mUniform.size() << LL_ENDL;
+            return ret;
+        }
         return mUniform[index];
     }
 
@@ -1436,6 +1481,8 @@ GLint LLGLSLShader::getUniformLocation(U32 index)
 
 GLint LLGLSLShader::getAttribLocation(U32 attrib)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+
     if (attrib < mAttribute.size())
     {
         return mAttribute[attrib];
@@ -1452,7 +1499,7 @@ void LLGLSLShader::uniform1i(const LLStaticHashedString& uniform, GLint v)
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(v,0.f,0.f,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec))
         {
@@ -1468,7 +1515,7 @@ void LLGLSLShader::uniform2i(const LLStaticHashedString& uniform, GLint i, GLint
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(i,j,0.f,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec))
         {
@@ -1485,7 +1532,7 @@ void LLGLSLShader::uniform1f(const LLStaticHashedString& uniform, GLfloat v)
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(v,0.f,0.f,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec))
         {
@@ -1501,7 +1548,7 @@ void LLGLSLShader::uniform2f(const LLStaticHashedString& uniform, GLfloat x, GLf
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(x,y,0.f,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec))
         {
@@ -1518,7 +1565,7 @@ void LLGLSLShader::uniform3f(const LLStaticHashedString& uniform, GLfloat x, GLf
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(x,y,z,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec))
         {
@@ -1534,7 +1581,7 @@ void LLGLSLShader::uniform1fv(const LLStaticHashedString& uniform, U32 count, co
 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(v[0],0.f,0.f,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
         {
@@ -1550,7 +1597,7 @@ void LLGLSLShader::uniform2fv(const LLStaticHashedString& uniform, U32 count, co
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(v[0],v[1],0.f,0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
         {
@@ -1566,7 +1613,7 @@ void LLGLSLShader::uniform3fv(const LLStaticHashedString& uniform, U32 count, co
                 
     if (location >= 0)
     {
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         LLVector4 vec(v[0],v[1],v[2],0.f);
         if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
         {
@@ -1583,12 +1630,11 @@ void LLGLSLShader::uniform4fv(const LLStaticHashedString& uniform, U32 count, co
     if (location >= 0)
     {
         LLVector4 vec(v);
-        std::map<GLint, LLVector4>::iterator iter = mValue.find(location);
+        const auto& iter = mValue.find(location);
         if (iter == mValue.end() || shouldChange(iter->second,vec) || count != 1)
         {
-            stop_glerror();
+            LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
             glUniform4fvARB(location, count, v);
-            stop_glerror();
             mValue[location] = vec;
         }
     }
@@ -1628,3 +1674,27 @@ void LLGLSLShader::setMinimumAlpha(F32 minimum)
     gGL.flush();
     uniform1f(LLShaderMgr::MINIMUM_ALPHA, minimum);
 }
+
+void LLShaderUniforms::apply(LLGLSLShader* shader)
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_SHADER;
+    for (auto& uniform : mIntegers)
+    {
+        shader->uniform1i(uniform.mUniform, uniform.mValue);
+    }
+
+    for (auto& uniform : mFloats)
+    {
+        shader->uniform1f(uniform.mUniform, uniform.mValue);
+    }
+
+    for (auto& uniform : mVectors)
+    {
+        shader->uniform4fv(uniform.mUniform, 1, uniform.mValue.mV);
+    }
+
+    for (auto& uniform : mVector3s)
+    {
+        shader->uniform3fv(uniform.mUniform, 1, uniform.mValue.mV);
+    }
+}
diff --git a/indra/llrender/llglslshader.h b/indra/llrender/llglslshader.h
index e2d0031740..83f11c050b 100644
--- a/indra/llrender/llglslshader.h
+++ b/indra/llrender/llglslshader.h
@@ -30,6 +30,7 @@
 #include "llgl.h"
 #include "llrender.h"
 #include "llstaticstringtable.h"
+#include <unordered_map>
 
 class LLShaderFeatures
 {
@@ -64,6 +65,66 @@ public:
 	LLShaderFeatures();
 };
 
+// ============= Structure for caching shader uniforms ===============
+class LLGLSLShader;
+
+class LLShaderUniforms
+{
+public:
+
+    template<typename T>
+    struct UniformSetting
+    {
+        S32 mUniform;
+        T mValue;
+    };
+
+    typedef UniformSetting<S32> IntSetting;
+    typedef UniformSetting<F32> FloatSetting;
+    typedef UniformSetting<LLVector4> VectorSetting;
+    typedef UniformSetting<LLVector3> Vector3Setting;
+
+    void clear()
+    {
+        mIntegers.resize(0);
+        mFloats.resize(0);
+        mVectors.resize(0);
+        mVector3s.resize(0);
+    }
+
+    void uniform1i(S32 index, S32 value)
+    {
+        mIntegers.push_back({ index, value });
+    }
+
+    void uniform1f(S32 index, F32 value)
+    {
+        mFloats.push_back({ index, value });
+    }
+
+    void uniform4fv(S32 index, const LLVector4& value)
+    {
+        mVectors.push_back({ index, value });
+    }
+    
+    void uniform4fv(S32 index, const F32* value)
+    {
+        mVectors.push_back({ index, LLVector4(value) });
+    }
+
+    void uniform3fv(S32 index, const LLVector3& value)
+    {
+        mVector3s.push_back({ index, value });
+    }
+
+    void apply(LLGLSLShader* shader);
+   
+
+    std::vector<IntSetting> mIntegers;
+    std::vector<FloatSetting> mFloats;
+    std::vector<VectorSetting> mVectors;
+    std::vector<Vector3Setting> mVector3s;
+};
 class LLGLSLShader
 {
 public:
@@ -75,12 +136,15 @@ public:
         , NUM_SHADER_CONSTS
     };
 
-	enum 
+    // enum primarily used to control application sky settings uniforms
+	typedef enum 
 	{
-		SG_DEFAULT = 0,
-		SG_SKY,
-		SG_WATER
-	};
+		SG_DEFAULT = 0,  // not sky or water specific
+		SG_SKY,  //
+		SG_WATER,
+        SG_ANY,
+        SG_COUNT
+	} eGroup;
 	
 	static std::set<LLGLSLShader*> sInstances;
 	static bool sProfileEnabled;
@@ -91,7 +155,6 @@ public:
 	static GLhandleARB sCurBoundShader;
 	static LLGLSLShader* sCurBoundShaderPtr;
 	static S32 sIndexedTextureChannels;
-	static bool sNoFixedFunction;
 
 	static void initProfile();
 	static void finishProfile(bool emit_report = true);
@@ -175,6 +238,8 @@ public:
 	
     BOOL link(BOOL suppress_errors = FALSE);
 	void bind();
+    //helper to conditionally bind mRiggedVariant instead of this
+    void bind(bool rigged);
 	void unbind();
 
 	// Unbinds any previously bound shader by explicitly binding no shader.
@@ -199,18 +264,21 @@ public:
 	U32 mAttributeMask;  //mask of which reserved attributes are set (lines up with LLVertexBuffer::getTypeMask())
 	std::vector<GLint> mUniform;   //lookup table of uniform enum to uniform location
 	LLStaticStringTable<GLint> mUniformMap; //lookup map of uniform name to uniform location
-	std::map<GLint, std::string> mUniformNameMap; //lookup map of uniform location to uniform name
-	std::map<GLint, LLVector4> mValue; //lookup map of uniform location to last known value
+    typedef std::unordered_map<GLint, std::string> uniform_name_map_t;
+    typedef std::unordered_map<GLint, LLVector4> uniform_value_map_t;
+    uniform_name_map_t mUniformNameMap; //lookup map of uniform location to uniform name
+	uniform_value_map_t mValue; //lookup map of uniform location to last known value
 	std::vector<GLint> mTexture;
 	S32 mTotalUniformSize;
 	S32 mActiveTextureChannels;
 	S32 mShaderLevel;
-	S32 mShaderGroup;
+	S32 mShaderGroup; // see LLGLSLShader::eGroup
 	BOOL mUniformsDirty;
 	LLShaderFeatures mFeatures;
 	std::vector< std::pair< std::string, GLenum > > mShaderFiles;
 	std::string mName;
-	boost::unordered_map<std::string, std::string> mDefines;
+    typedef std::unordered_map<std::string, std::string> defines_map_t;
+	defines_map_t mDefines;
 
 	//statistcis for profiling shader performance
 	U32 mTimerQuery;
@@ -228,6 +296,9 @@ public:
 	std::vector<U32> mTextureMagFilter;
 	std::vector<U32> mTextureMinFilter;
 
+    // this pointer should be set to whichever shader represents this shader's rigged variant
+    LLGLSLShader* mRiggedVariant = nullptr;
+
 private:
 	void unloadInternal();
 };
diff --git a/indra/llrender/llgltexture.cpp b/indra/llrender/llgltexture.cpp
index ad501687ed..e012eb9a62 100644
--- a/indra/llrender/llgltexture.cpp
+++ b/indra/llrender/llgltexture.cpp
@@ -164,11 +164,11 @@ BOOL LLGLTexture::createGLTexture()
 	return mGLTexturep->createGLTexture() ;
 }
 
-BOOL LLGLTexture::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename, BOOL to_create, S32 category)
+BOOL LLGLTexture::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename, BOOL to_create, S32 category, bool defer_copy, LLGLuint* tex_name)
 {
-	llassert(mGLTexturep.notNull()) ;	
+	llassert(mGLTexturep.notNull());
 
-	BOOL ret = mGLTexturep->createGLTexture(discard_level, imageraw, usename, to_create, category) ;
+	BOOL ret = mGLTexturep->createGLTexture(discard_level, imageraw, usename, to_create, category, defer_copy, tex_name) ;
 
 	if(ret)
 	{
@@ -260,18 +260,20 @@ LLTexUnit::eTextureType LLGLTexture::getTarget(void) const
 	return mGLTexturep->getTarget() ;
 }
 
-BOOL LLGLTexture::setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height)
+BOOL LLGLTexture::setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height, LLGLuint use_name)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	llassert(mGLTexturep.notNull()) ;
 
-	return mGLTexturep->setSubImage(imageraw, x_pos, y_pos, width, height) ;
+	return mGLTexturep->setSubImage(imageraw, x_pos, y_pos, width, height, 0, use_name) ;
 }
 
-BOOL LLGLTexture::setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height)
+BOOL LLGLTexture::setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height, LLGLuint use_name)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	llassert(mGLTexturep.notNull()) ;
 
-	return mGLTexturep->setSubImage(datap, data_width, data_height, x_pos, y_pos, width, height) ;
+	return mGLTexturep->setSubImage(datap, data_width, data_height, x_pos, y_pos, width, height, 0, use_name) ;
 }
 
 void LLGLTexture::setGLTextureCreated (bool initialized)
diff --git a/indra/llrender/llgltexture.h b/indra/llrender/llgltexture.h
index 071912c2c2..8cfe7b62de 100644
--- a/indra/llrender/llgltexture.h
+++ b/indra/llrender/llgltexture.h
@@ -124,13 +124,22 @@ public:
 	BOOL       hasGLTexture() const ;
 	LLGLuint   getTexName() const ;		
 	BOOL       createGLTexture() ;
-	BOOL       createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename = 0, BOOL to_create = TRUE, S32 category = LLGLTexture::OTHER);
+	
+    // Create a GL Texture from an image raw
+    // discard_level - mip level, 0 for highest resultion mip
+    // imageraw - the image to copy from
+    // usename - explicit GL name override
+    // to_create - set to FALSE to force gl texture to not be created
+    // category - LLGLTexture category for this LLGLTexture
+    // defer_copy - set to true to allocate GL texture but NOT initialize with imageraw data
+    // tex_name - if not null, will be set to the GL name of the texture created
+    BOOL       createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename = 0, BOOL to_create = TRUE, S32 category = LLGLTexture::OTHER, bool defer_copy = false, LLGLuint* tex_name = nullptr);
 
 	void       setFilteringOption(LLTexUnit::eTextureFilterOptions option);
 	void       setExplicitFormat(LLGLint internal_format, LLGLenum primary_format, LLGLenum type_format = 0, BOOL swap_bytes = FALSE);
 	void       setAddressMode(LLTexUnit::eTextureAddressMode mode);
-	BOOL       setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height);
-	BOOL       setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height);
+	BOOL       setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height, LLGLuint use_name = 0);
+	BOOL       setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height, LLGLuint use_name = 0);
 	void       setGLTextureCreated (bool initialized);
 	void       setCategory(S32 category) ;
     void       setTexName(LLGLuint); // for forcing w/ externally created textures only
@@ -176,7 +185,7 @@ private:
 protected:
 	void setTexelsPerImage();
 
-	//note: do not make this function public.
+public:
 	/*virtual*/ LLImageGL* getGLTexture() const ;
 
 protected:
diff --git a/indra/llrender/llimagegl.cpp b/indra/llrender/llimagegl.cpp
index 0151d20128..9bd3a0a6b0 100644
--- a/indra/llrender/llimagegl.cpp
+++ b/indra/llrender/llimagegl.cpp
@@ -39,6 +39,12 @@
 #include "llgl.h"
 #include "llglslshader.h"
 #include "llrender.h"
+#include "llwindow.h"
+#include "llframetimer.h"
+
+#if !LL_IMAGEGL_THREAD_CHECK
+#define checkActiveThread()
+#endif
 
 //----------------------------------------------------------------------------
 const F32 MIN_TEXTURE_LIFETIME = 10.f;
@@ -61,9 +67,11 @@ F32 LLImageGL::sLastFrameTime			= 0.f;
 BOOL LLImageGL::sAllowReadBackRaw       = FALSE ;
 LLImageGL* LLImageGL::sDefaultGLTexture = NULL ;
 bool LLImageGL::sCompressTextures = false;
-
 std::set<LLImageGL*> LLImageGL::sImageList;
 
+
+bool LLImageGLThread::sEnabled = false;
+
 //****************************************************************************************************
 //The below for texture auditing use only
 //****************************************************************************************************
@@ -170,15 +178,24 @@ BOOL is_little_endian()
     
 	return (*c == 0x78) ;
 }
+
 //static 
-void LLImageGL::initClass(S32 num_catagories, BOOL skip_analyze_alpha /* = false */)
+void LLImageGL::initClass(LLWindow* window, S32 num_catagories, BOOL skip_analyze_alpha /* = false */, bool multi_threaded /* = false */)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	sSkipAnalyzeAlpha = skip_analyze_alpha;
+
+    if (multi_threaded)
+    {
+        LLImageGLThread::createInstance(window);
+    }
 }
 
 //static 
 void LLImageGL::cleanupClass() 
-{	
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    LLImageGLThread::deleteSingleton();
 }
 
 //static
@@ -194,6 +211,7 @@ S32 LLImageGL::dataFormatBits(S32 dataformat)
     case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:    return 8;
     case GL_LUMINANCE:						        return 8;
     case GL_ALPHA:							        return 8;
+    case GL_RED:                                    return 8;
     case GL_COLOR_INDEX:						    return 8;
     case GL_LUMINANCE_ALPHA:					    return 16;
     case GL_RGB:								    return 24;
@@ -243,6 +261,7 @@ S32 LLImageGL::dataFormatComponents(S32 dataformat)
 	  case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT: return 4;
 	  case GL_LUMINANCE:						return 1;
 	  case GL_ALPHA:							return 1;
+      case GL_RED:                              return 1;
 	  case GL_COLOR_INDEX:						return 1;
 	  case GL_LUMINANCE_ALPHA:					return 2;
 	  case GL_RGB:								return 3;
@@ -258,11 +277,10 @@ S32 LLImageGL::dataFormatComponents(S32 dataformat)
 
 //----------------------------------------------------------------------------
 
-static LLTrace::BlockTimerStatHandle FTM_IMAGE_UPDATE_STATS("Image Stats");
 // static
 void LLImageGL::updateStats(F32 current_time)
 {
-	LL_RECORD_BLOCK_TIME(FTM_IMAGE_UPDATE_STATS);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	sLastFrameTime = current_time;
 	sBoundTextureMemory = sCurBoundTextureMemory;
 	sCurBoundTextureMemory = S32Bytes(0);
@@ -295,10 +313,8 @@ void LLImageGL::destroyGL(BOOL save_state)
 			if (save_state && glimage->isGLTextureCreated() && glimage->mComponents)
 			{
 				glimage->mSaveData = new LLImageRaw;
-				glimage->claimMem(glimage->mSaveData);
 				if(!glimage->readBackRaw(glimage->mCurrentDiscardLevel, glimage->mSaveData, false)) //necessary, keep it.
 				{
-					glimage->disclaimMem(glimage->mSaveData);
 					glimage->mSaveData = NULL ;
 				}
 			}
@@ -372,8 +388,7 @@ BOOL LLImageGL::create(LLPointer<LLImageGL>& dest, const LLImageRaw* imageraw, B
 //----------------------------------------------------------------------------
 
 LLImageGL::LLImageGL(BOOL usemipmaps)
-:	LLTrace::MemTrackable<LLImageGL>("LLImageGL"),
-    mSaveData(0), mExternalTexture(FALSE)
+:	mSaveData(0), mExternalTexture(FALSE)
 {
 	init(usemipmaps);
 	setSize(0, 0, 0);
@@ -382,8 +397,7 @@ LLImageGL::LLImageGL(BOOL usemipmaps)
 }
 
 LLImageGL::LLImageGL(U32 width, U32 height, U8 components, BOOL usemipmaps)
-:	LLTrace::MemTrackable<LLImageGL>("LLImageGL"),
-    mSaveData(0), mExternalTexture(FALSE)
+:	mSaveData(0), mExternalTexture(FALSE)
 {
 	llassert( components <= 4 );
 	init(usemipmaps);
@@ -393,8 +407,7 @@ LLImageGL::LLImageGL(U32 width, U32 height, U8 components, BOOL usemipmaps)
 }
 
 LLImageGL::LLImageGL(const LLImageRaw* imageraw, BOOL usemipmaps)
-:	LLTrace::MemTrackable<LLImageGL>("LLImageGL"),
-    mSaveData(0), mExternalTexture(FALSE)
+:	mSaveData(0), mExternalTexture(FALSE)
 {
 	init(usemipmaps);
 	setSize(0, 0, 0);
@@ -412,7 +425,6 @@ LLImageGL::LLImageGL(
     LLGLenum formatPrimary,
     LLGLenum formatType,
     LLTexUnit::eTextureAddressMode addressMode)
-    : LLTrace::MemTrackable<LLImageGL>("LLImageGL"), mSaveData(0), mExternalTexture(TRUE)
 {
     init(false);
     mTexName = texName;
@@ -427,7 +439,7 @@ LLImageGL::LLImageGL(
 
 LLImageGL::~LLImageGL()
 {
-    if (!mExternalTexture)
+    if (!mExternalTexture && gGLManager.mInited)
     {
 	    LLImageGL::cleanup();
 	    sImageList.erase(this);
@@ -438,6 +450,10 @@ LLImageGL::~LLImageGL()
 
 void LLImageGL::init(BOOL usemipmaps)
 {
+#if LL_IMAGEGL_THREAD_CHECK
+    mActiveThread = LLThread::currentID();
+#endif
+
 	// keep these members in the same order as declared in llimagehl.h
 	// so that it is obvious by visual inspection if we forgot to
 	// init a field.
@@ -493,6 +509,9 @@ void LLImageGL::init(BOOL usemipmaps)
 #endif
 
 	mCategory = -1;
+
+	// Sometimes we have to post work for the main thread.
+	mMainQueue = LL::WorkQueue::getInstance("mainloop");
 }
 
 void LLImageGL::cleanup()
@@ -534,18 +553,12 @@ bool LLImageGL::setSize(S32 width, S32 height, S32 ncomponents, S32 discard_leve
 	if (width != mWidth || height != mHeight || ncomponents != mComponents)
 	{
 		// Check if dimensions are a power of two!
-		if (!checkSize(width,height))
+		if (!checkSize(width, height))
 		{
 			LL_WARNS() << llformat("Texture has non power of two dimension: %dx%d",width,height) << LL_ENDL;
 			return false;
 		}
 		
-		if (mTexName)
-		{
-// 			LL_WARNS() << "Setting Size of LLImageGL with existing mTexName = " << mTexName << LL_ENDL;
-			destroyGLTexture();
-		}
-
 		// pickmask validity depends on old image size, delete it
 		freePickMask();
 
@@ -656,6 +669,7 @@ void LLImageGL::setExplicitFormat( LLGLint internal_format, LLGLenum primary_for
 
 void LLImageGL::setImage(const LLImageRaw* imageraw)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	llassert((imageraw->getWidth() == getWidth(mCurrentDiscardLevel)) &&
 			 (imageraw->getHeight() == getHeight(mCurrentDiscardLevel)) &&
 			 (imageraw->getComponents() == getComponents()));
@@ -663,10 +677,9 @@ void LLImageGL::setImage(const LLImageRaw* imageraw)
 	setImage(rawdata, FALSE);
 }
 
-static LLTrace::BlockTimerStatHandle FTM_SET_IMAGE("setImage");
-BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
+BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips /* = FALSE */, S32 usename /* = 0 */)
 {
-	LL_RECORD_BLOCK_TIME(FTM_SET_IMAGE);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	bool is_compressed = false;
 
     switch (mFormatPrimary)
@@ -683,12 +696,11 @@ BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
         break;
     }
 	
-	
-	
 	if (mUseMipMaps)
 	{
 		//set has mip maps to true before binding image so tex parameters get set properly
-		gGL.getTexUnit(0)->unbind(mBindTarget);
+        gGL.getTexUnit(0)->unbind(mBindTarget);
+        
 		mHasMipMaps = true;
 		mTexOptionsDirty = true;
 		setFilteringOption(LLTexUnit::TFO_ANISOTROPIC);
@@ -698,10 +710,16 @@ BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 		mHasMipMaps = false;
 	}
 	
-	llverify(gGL.getTexUnit(0)->bind(this));
-	
-	
-	if (mUseMipMaps)
+    gGL.getTexUnit(0)->bind(this, false, false, usename);
+
+    if (data_in == nullptr)
+    {
+        S32 w = getWidth();
+        S32 h = getHeight();
+        LLImageGL::setManualImage(mTarget, 0, mFormatInternal, w, h,
+            mFormatPrimary, mFormatType, (GLvoid*)data_in, mAllowCompression);
+    }
+    else if (mUseMipMaps)
 	{
 		if (data_hasmips)
 		{
@@ -728,8 +746,6 @@ BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 				}
 				else
 				{
-// 					LL_RECORD_BLOCK_TIME(FTM_TEMP4);
-
 					if(mFormatSwapBytes)
 					{
 						glPixelStorei(GL_UNPACK_SWAP_BYTES, 1);
@@ -760,8 +776,6 @@ BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 			{
 				stop_glerror();
 				{
-// 					LL_RECORD_BLOCK_TIME(FTM_TEMP4);
-
 					if(mFormatSwapBytes)
 					{
 						glPixelStorei(GL_UNPACK_SWAP_BYTES, 1);
@@ -781,7 +795,7 @@ BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 						glTexParameteri(mTarget, GL_GENERATE_MIPMAP, GL_TRUE);
 					}
 
-					LLImageGL::setManualImage(mTarget, 0, mFormatInternal,
+                    LLImageGL::setManualImage(mTarget, 0, mFormatInternal,
 								 w, h, 
 								 mFormatPrimary, mFormatType,
 								 data_in, mAllowCompression);
@@ -871,14 +885,13 @@ BOOL LLImageGL::setImage(const U8* data_in, BOOL data_hasmips)
 					llassert(w > 0 && h > 0 && cur_mip_data);
 					(void)cur_mip_data;
 					{
-// 						LL_RECORD_BLOCK_TIME(FTM_TEMP4);
 						if(mFormatSwapBytes)
 						{
 							glPixelStorei(GL_UNPACK_SWAP_BYTES, 1);
 							stop_glerror();
 						}
 
-						LLImageGL::setManualImage(mTarget, m, mFormatInternal, w, h, mFormatPrimary, mFormatType, cur_mip_data, mAllowCompression);
+                        LLImageGL::setManualImage(mTarget, m, mFormatInternal, w, h, mFormatPrimary, mFormatType, cur_mip_data, mAllowCompression);
 						if (m == 0)
 						{
 							analyzeAlpha(data_in, w, h);
@@ -1065,13 +1078,15 @@ void LLImageGL::postAddToAtlas()
 	stop_glerror();	
 }
 
-BOOL LLImageGL::setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update)
+BOOL LLImageGL::setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update /* = FALSE */, LLGLuint use_name)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
 	if (!width || !height)
 	{
 		return TRUE;
 	}
-	if (mTexName == 0)
+    LLGLuint tex_name = use_name != 0 ? use_name : mTexName;
+	if (0 == tex_name)
 	{
 		// *TODO: Re-enable warning?  Ran into thread locking issues? DK 2011-02-18
 		//LL_WARNS() << "Setting subimage on image without GL texture" << LL_ENDL;
@@ -1087,7 +1102,7 @@ BOOL LLImageGL::setSubImage(const U8* datap, S32 data_width, S32 data_height, S3
 	// HACK: allow the caller to explicitly force the fast path (i.e. using glTexSubImage2D here instead of calling setImage) even when updating the full texture.
 	if (!force_fast_update && x_pos == 0 && y_pos == 0 && width == getWidth() && height == getHeight() && data_width == width && data_height == height)
 	{
-		setImage(datap, FALSE);
+		setImage(datap, FALSE, tex_name);
 	}
 	else
 	{
@@ -1139,12 +1154,11 @@ BOOL LLImageGL::setSubImage(const U8* datap, S32 data_width, S32 data_height, S3
 
 		datap += (y_pos * data_width + x_pos) * getComponents();
 		// Update the GL texture
-		BOOL res = gGL.getTexUnit(0)->bindManual(mBindTarget, mTexName);
+		BOOL res = gGL.getTexUnit(0)->bindManual(mBindTarget, tex_name);
 		if (!res) LL_ERRS() << "LLImageGL::setSubImage(): bindTexture failed" << LL_ENDL;
 		stop_glerror();
 
-		glTexSubImage2D(mTarget, 0, x_pos, y_pos, 
-						width, height, mFormatPrimary, mFormatType, datap);
+		glTexSubImage2D(mTarget, 0, x_pos, y_pos, width, height, mFormatPrimary, mFormatType, datap);
 		gGL.getTexUnit(0)->disable();
 		stop_glerror();
 
@@ -1161,9 +1175,10 @@ BOOL LLImageGL::setSubImage(const U8* datap, S32 data_width, S32 data_height, S3
 	return TRUE;
 }
 
-BOOL LLImageGL::setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update)
+BOOL LLImageGL::setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update /* = FALSE */, LLGLuint use_name)
 {
-	return setSubImage(imageraw->getData(), imageraw->getWidth(), imageraw->getHeight(), x_pos, y_pos, width, height, force_fast_update);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+	return setSubImage(imageraw->getData(), imageraw->getWidth(), imageraw->getHeight(), x_pos, y_pos, width, height, force_fast_update, use_name);
 }
 
 // Copy sub image from frame buffer
@@ -1183,15 +1198,36 @@ BOOL LLImageGL::setSubImageFromFrameBuffer(S32 fb_x, S32 fb_y, S32 x_pos, S32 y_
 }
 
 // static
-static LLTrace::BlockTimerStatHandle FTM_GENERATE_TEXTURES("generate textures");
 void LLImageGL::generateTextures(S32 numTextures, U32 *textures)
 {
-	LL_RECORD_BLOCK_TIME(FTM_GENERATE_TEXTURES);
-	glGenTextures(numTextures, textures);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    static constexpr U32 pool_size = 1024;
+    static thread_local U32 name_pool[pool_size]; // pool of texture names
+    static thread_local U32 name_count = 0; // number of available names in the pool
+
+    if (name_count == 0)
+    {
+        LL_PROFILE_ZONE_NAMED("iglgt - reup pool");
+        // pool is emtpy, refill it
+        glGenTextures(pool_size, name_pool);
+        name_count = pool_size;
+    }
+
+    if (numTextures <= name_count)
+    {
+        //copy teture names off the end of the pool
+        memcpy(textures, name_pool + name_count - numTextures, sizeof(U32) * numTextures);
+        name_count -= numTextures;
+    }
+    else
+    {
+        LL_PROFILE_ZONE_NAMED("iglgt - pool miss");
+        glGenTextures(numTextures, textures);
+    }
 }
 
 // static
-void LLImageGL::deleteTextures(S32 numTextures, U32 *textures)
+void LLImageGL::deleteTextures(S32 numTextures, const U32 *textures)
 {
 	if (gGLManager.mInited)
 	{
@@ -1200,125 +1236,140 @@ void LLImageGL::deleteTextures(S32 numTextures, U32 *textures)
 }
 
 // static
-static LLTrace::BlockTimerStatHandle FTM_SET_MANUAL_IMAGE("setManualImage");
-void LLImageGL::setManualImage(U32 target, S32 miplevel, S32 intformat, S32 width, S32 height, U32 pixformat, U32 pixtype, const void *pixels, bool allow_compression)
+void LLImageGL::setManualImage(U32 target, S32 miplevel, S32 intformat, S32 width, S32 height, U32 pixformat, U32 pixtype, const void* pixels, bool allow_compression)
 {
-	LL_RECORD_BLOCK_TIME(FTM_SET_MANUAL_IMAGE);
-	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];
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    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
+            if (pixels != nullptr)
+            {
+                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;
-		}
+                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];
+                }
+            }
 
-		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];
+            pixformat = GL_RGBA;
+            intformat = GL_RGBA8;
+        }
 
-			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];
+        if (pixformat == GL_LUMINANCE_ALPHA && pixtype == GL_UNSIGNED_BYTE)
+        { //GL_LUMINANCE_ALPHA is deprecated, convert to RGBA
+            if (pixels != nullptr)
+            {
+                use_scratch = true;
+                scratch = new U32[width * height];
 
-				U8* pix = (U8*) &scratch[i];
-				pix[0] = pix[1] = pix[2] = lum;
-				pix[3] = alpha;
-			}				
-			
-			pixformat = GL_RGBA;
-			intformat = GL_RGBA8;
-		}
+                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];
 
-		if (pixformat == GL_LUMINANCE && pixtype == GL_UNSIGNED_BYTE) 
-		{ //GL_LUMINANCE_ALPHA is deprecated, convert to RGB
-			use_scratch = true;
-			scratch = new U32[width*height];
+                    U8* pix = (U8*)&scratch[i];
+                    pix[0] = pix[1] = pix[2] = lum;
+                    pix[3] = alpha;
+                }
+            }
 
-			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;
-		}
-	}
+            pixformat = GL_RGBA;
+            intformat = GL_RGBA8;
+        }
 
-	if (LLImageGL::sCompressTextures && allow_compression)
-	{
-		switch (intformat)
-		{
-			case GL_RGB: 
-			case GL_RGB8:
-				intformat = GL_COMPRESSED_RGB; 
-				break;
-            case GL_SRGB:
-            case GL_SRGB8:
-                intformat = GL_COMPRESSED_SRGB;
-                break;
-			case GL_RGBA:
-			case GL_RGBA8:
-				intformat = GL_COMPRESSED_RGBA; 
-				break;
-            case GL_SRGB_ALPHA:
-            case GL_SRGB8_ALPHA8:
-                intformat = GL_COMPRESSED_SRGB_ALPHA;
-                break;
-			case GL_LUMINANCE:
-			case GL_LUMINANCE8:
-				intformat = GL_COMPRESSED_LUMINANCE;
-				break;
-			case GL_LUMINANCE_ALPHA:
-			case GL_LUMINANCE8_ALPHA8:
-				intformat = GL_COMPRESSED_LUMINANCE_ALPHA;
-				break;
-			case GL_ALPHA:
-			case GL_ALPHA8:
-				intformat = GL_COMPRESSED_ALPHA;
-				break;
-			default:
-				LL_WARNS() << "Could not compress format: " << std::hex << intformat << LL_ENDL;
-				break;
-		}
-	}
+        if (pixformat == GL_LUMINANCE && pixtype == GL_UNSIGNED_BYTE)
+        { //GL_LUMINANCE_ALPHA is deprecated, convert to RGB
+            if (pixels != nullptr)
+            {
+                use_scratch = true;
+                scratch = new U32[width * height];
 
-	stop_glerror();
-	glTexImage2D(target, miplevel, intformat, width, height, 0, pixformat, pixtype, use_scratch ? scratch : pixels);
-	stop_glerror();
+                U32 pixel_count = (U32)(width * height);
+                for (U32 i = 0; i < pixel_count; i++)
+                {
+                    U8 lum = ((U8*)pixels)[i];
 
-	if (use_scratch)
-	{
-		delete [] scratch;
-	}
+                    U8* pix = (U8*)&scratch[i];
+                    pix[0] = pix[1] = pix[2] = lum;
+                    pix[3] = 255;
+                }
+            }
+            pixformat = GL_RGBA;
+            intformat = GL_RGB8;
+        }
+    }
+
+    if (LLImageGL::sCompressTextures && allow_compression)
+    {
+        switch (intformat)
+        {
+        case GL_RGB:
+        case GL_RGB8:
+            intformat = GL_COMPRESSED_RGB;
+            break;
+        case GL_SRGB:
+        case GL_SRGB8:
+            intformat = GL_COMPRESSED_SRGB;
+            break;
+        case GL_RGBA:
+        case GL_RGBA8:
+            intformat = GL_COMPRESSED_RGBA;
+            break;
+        case GL_SRGB_ALPHA:
+        case GL_SRGB8_ALPHA8:
+            intformat = GL_COMPRESSED_SRGB_ALPHA;
+            break;
+        case GL_LUMINANCE:
+        case GL_LUMINANCE8:
+            intformat = GL_COMPRESSED_LUMINANCE;
+            break;
+        case GL_LUMINANCE_ALPHA:
+        case GL_LUMINANCE8_ALPHA8:
+            intformat = GL_COMPRESSED_LUMINANCE_ALPHA;
+            break;
+        case GL_ALPHA:
+        case GL_ALPHA8:
+            intformat = GL_COMPRESSED_ALPHA;
+            break;
+        case GL_RED:
+        case GL_R8:
+            intformat = GL_COMPRESSED_RED;
+            break;
+        default:
+            LL_WARNS() << "Could not compress format: " << std::hex << intformat << LL_ENDL;
+            break;
+        }
+    }
+
+    stop_glerror();
+    {
+        LL_PROFILE_ZONE_NAMED("glTexImage2D");
+        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
 //the texture is assiciate with some image by calling glTexImage outside LLImageGL
-static LLTrace::BlockTimerStatHandle FTM_CREATE_GL_TEXTURE1("createGLTexture()");
 BOOL LLImageGL::createGLTexture()
 {
-	LL_RECORD_BLOCK_TIME(FTM_CREATE_GL_TEXTURE1);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    checkActiveThread();
+
 	if (gGLManager.mIsDisabled)
 	{
 		LL_WARNS() << "Trying to create a texture while GL is disabled!" << LL_ENDL;
@@ -1333,6 +1384,7 @@ BOOL LLImageGL::createGLTexture()
 	if(mTexName)
 	{
 		LLImageGL::deleteTextures(1, (reinterpret_cast<GLuint*>(&mTexName))) ;
+        mTexName = 0;
 	}
 	
 
@@ -1347,23 +1399,24 @@ BOOL LLImageGL::createGLTexture()
 	return TRUE ;
 }
 
-static LLTrace::BlockTimerStatHandle FTM_CREATE_GL_TEXTURE2("createGLTexture(raw)");
-BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename/*=0*/, BOOL to_create, S32 category)
+BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename/*=0*/, BOOL to_create, S32 category, bool defer_copy, LLGLuint* tex_name)
 {
-	LL_RECORD_BLOCK_TIME(FTM_CREATE_GL_TEXTURE2);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    checkActiveThread();
+
 	if (gGLManager.mIsDisabled)
 	{
 		LL_WARNS() << "Trying to create a texture while GL is disabled!" << LL_ENDL;
 		return FALSE;
 	}
 
-	mGLTextureCreated = false ;
 	llassert(gGLManager.mInited);
 	stop_glerror();
 
 	if (!imageraw || imageraw->isBufferInvalid())
 	{
 		LL_WARNS() << "Trying to create a texture from invalid image data" << LL_ENDL;
+        mGLTextureCreated = false;
 		return FALSE;
 	}
 
@@ -1376,6 +1429,7 @@ BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S
 	// Actual image width/height = raw image width/height * 2^discard_level
 	S32 raw_w = imageraw->getWidth() ;
 	S32 raw_h = imageraw->getHeight() ;
+
 	S32 w = raw_w << discard_level;
 	S32 h = raw_h << discard_level;
 
@@ -1383,6 +1437,7 @@ BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S
 	if (!setSize(w, h, imageraw->getComponents(), discard_level))
 	{
 		LL_WARNS() << "Trying to create a texture with incorrect dimensions!" << LL_ENDL;
+        mGLTextureCreated = false;
 		return FALSE;
 	}
 
@@ -1451,115 +1506,236 @@ BOOL LLImageGL::createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S
 		destroyGLTexture();
 		mCurrentDiscardLevel = discard_level;	
 		mLastBindTime = sLastFrameTime;
+        mGLTextureCreated = false;
 		return TRUE ;
 	}
 
 	setCategory(category);
  	const U8* rawdata = imageraw->getData();
-	return createGLTexture(discard_level, rawdata, FALSE, usename);
+	return createGLTexture(discard_level, rawdata, FALSE, usename, defer_copy, tex_name);
 }
 
-static LLTrace::BlockTimerStatHandle FTM_CREATE_GL_TEXTURE3("createGLTexture3(data)");
-BOOL LLImageGL::createGLTexture(S32 discard_level, const U8* data_in, BOOL data_hasmips, S32 usename)
+BOOL LLImageGL::createGLTexture(S32 discard_level, const U8* data_in, BOOL data_hasmips, S32 usename, bool defer_copy, LLGLuint* tex_name)
+// Call with void data, vmem is allocated but unitialized
 {
-	LL_RECORD_BLOCK_TIME(FTM_CREATE_GL_TEXTURE3);
-	llassert(data_in);
-	stop_glerror();
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    checkActiveThread();
 
-	if (discard_level < 0)
-	{
-		llassert(mCurrentDiscardLevel >= 0);
-		discard_level = mCurrentDiscardLevel;
-	}
-	discard_level = llclamp(discard_level, 0, (S32)mMaxDiscardLevel);
+    bool main_thread = on_main_thread();
 
-	if (mTexName != 0 && discard_level == mCurrentDiscardLevel)
-	{
-		// This will only be true if the size has not changed
-		return setImage(data_in, data_hasmips);
-	}
-	
-	U32 old_name = mTexName;
-// 	S32 old_discard = mCurrentDiscardLevel;
-	
-	if (usename != 0)
-	{
-		mTexName = usename;
-	}
-	else
-	{
-		LLImageGL::generateTextures(1, &mTexName);
-		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)
-	{
-		if (old_name)
-		{
-			sGlobalTextureMemory -= mTextureMemory;
-			LLImageGL::deleteTextures(1, &old_name);
-			disclaimMem(mTextureMemory);
-			stop_glerror();
-		}
+    if (defer_copy)
+    {
+        data_in = nullptr;
+    }
+    else
+    {
+        llassert(data_in);
+    }
 
-		LL_WARNS() << "LLImageGL::createGLTexture failed to make texture" << LL_ENDL;
-		return FALSE;
-	}
+    stop_glerror();
 
-	if (mUseMipMaps)
-	{
-		mAutoGenMips = gGLManager.mHasMipMapGeneration;
-#if LL_DARWIN
-		// On the Mac GF2 and GF4MX drivers, auto mipmap generation doesn't work right with alpha-only textures.
-		if(gGLManager.mIsGF2or4MX && (mFormatInternal == GL_ALPHA8) && (mFormatPrimary == GL_ALPHA))
-		{
-			mAutoGenMips = FALSE;
-		}
-#endif
-	}
+    if (discard_level < 0)
+    {
+        llassert(mCurrentDiscardLevel >= 0);
+        discard_level = mCurrentDiscardLevel;
+    }
+    discard_level = llclamp(discard_level, 0, (S32)mMaxDiscardLevel);
 
-	mCurrentDiscardLevel = discard_level;	
+    if (main_thread // <--- always force creation of new_texname when not on main thread ...
+        && !defer_copy // <--- ... or defer copy is set
+        && mTexName != 0 && discard_level == mCurrentDiscardLevel)
+    {
+        LL_PROFILE_ZONE_NAMED("cglt - early setImage");
+        // This will only be true if the size has not changed
+        if (tex_name != nullptr)
+        {
+            *tex_name = mTexName;
+        }
+        return setImage(data_in, data_hasmips);
+    }
 
-	if (!setImage(data_in, data_hasmips))
-	{
-		stop_glerror();
-		return FALSE;
-	}
+    GLuint old_texname = mTexName;
+    GLuint new_texname = 0;
+    if (usename != 0)
+    {
+        llassert(main_thread);
+        new_texname = usename;
+    }
+    else
+    {
+        LLImageGL::generateTextures(1, &new_texname);
+        {
+            gGL.getTexUnit(0)->bind(this, false, false, new_texname);
+            glTexParameteri(LLTexUnit::getInternalType(mBindTarget), GL_TEXTURE_BASE_LEVEL, 0);
+            glTexParameteri(LLTexUnit::getInternalType(mBindTarget), GL_TEXTURE_MAX_LEVEL, mMaxDiscardLevel - discard_level);
+        }
+    }
+
+    if (tex_name != nullptr)
+    {
+        *tex_name = new_texname;
+    }
 
-	// Set texture options to our defaults.
-	gGL.getTexUnit(0)->setHasMipMaps(mHasMipMaps);
-	gGL.getTexUnit(0)->setTextureAddressMode(mAddressMode);
-	gGL.getTexUnit(0)->setTextureFilteringOption(mFilterOption);
+    if (mUseMipMaps)
+    {
+        mAutoGenMips = gGLManager.mHasMipMapGeneration;
+    }
 
-	// things will break if we don't unbind after creation
-	gGL.getTexUnit(0)->unbind(mBindTarget);
-	stop_glerror();
+    mCurrentDiscardLevel = discard_level;
 
-	if (old_name != 0)
-	{
-		sGlobalTextureMemory -= mTextureMemory;
+    {
+        LL_PROFILE_ZONE_NAMED("cglt - late setImage");
+        if (!setImage(data_in, data_hasmips, new_texname))
+        {
+            return FALSE;
+        }
+    }
 
-		LLImageGL::deleteTextures(1, &old_name);
+    // Set texture options to our defaults.
+    gGL.getTexUnit(0)->setHasMipMaps(mHasMipMaps);
+    gGL.getTexUnit(0)->setTextureAddressMode(mAddressMode);
+    gGL.getTexUnit(0)->setTextureFilteringOption(mFilterOption);
 
-		stop_glerror();
-	}
+    // things will break if we don't unbind after creation
+    gGL.getTexUnit(0)->unbind(mBindTarget);
 
-	disclaimMem(mTextureMemory);
-	mTextureMemory = (S32Bytes)getMipBytes(discard_level);
-	claimMem(mTextureMemory);
-	sGlobalTextureMemory += mTextureMemory;
-	mTexelsInGLTexture = getWidth() * getHeight() ;
+    if (old_texname != 0)
+    {
+        sGlobalTextureMemory -= mTextureMemory;
+    }
 
-	// mark this as bound at this point, so we don't throw it out immediately
-	mLastBindTime = sLastFrameTime;
-	return TRUE;
+    //if we're on the image loading thread, be sure to delete old_texname and update mTexName on the main thread
+    if (!defer_copy)
+    {
+        if (!main_thread)
+        {
+            syncToMainThread(new_texname);
+        }
+        else
+        {
+            //not on background thread, immediately set mTexName
+            if (old_texname != 0 && old_texname != new_texname)
+            {
+                LLImageGL::deleteTextures(1, &old_texname);
+            }
+            mTexName = new_texname;
+        }
+    }
+
+    
+    mTextureMemory = (S32Bytes)getMipBytes(mCurrentDiscardLevel);
+    sGlobalTextureMemory += mTextureMemory;
+    mTexelsInGLTexture = getWidth() * getHeight();
+
+    // mark this as bound at this point, so we don't throw it out immediately
+    mLastBindTime = sLastFrameTime;
+
+    checkActiveThread();
+    return TRUE;
+}
+
+void LLImageGLThread::updateClass()
+{
+    LL_PROFILE_ZONE_SCOPED;
+
+    // update available vram one per second
+    static LLFrameTimer sTimer;
+
+    if (sTimer.getElapsedSeconds() < 1.f)
+    {
+        return;
+    }
+    
+    sTimer.reset();
+
+    auto func = []()
+    {
+        if (gGLManager.mHasATIMemInfo)
+        {
+            S32 meminfo[4];
+            glGetIntegerv(GL_TEXTURE_FREE_MEMORY_ATI, meminfo);
+            LLImageGLThread::sFreeVRAMMegabytes = meminfo[0];
+
+        }
+        else if (gGLManager.mHasNVXMemInfo)
+        {
+            S32 free_memory;
+            glGetIntegerv(GL_GPU_MEMORY_INFO_CURRENT_AVAILABLE_VIDMEM_NVX, &free_memory);
+            LLImageGLThread::sFreeVRAMMegabytes = free_memory / 1024;
+        }
+    };
+
+    
+    // post update to background thread if available, otherwise execute immediately
+    auto queue = LL::WorkQueue::getInstance("LLImageGL");
+    if (sEnabled)
+    {
+        queue->post(func);
+    }
+    else
+    {
+        llassert(queue == nullptr);
+        func();
+    }
+}
+
+void LLImageGL::syncToMainThread(LLGLuint new_tex_name)
+{
+    LL_PROFILE_ZONE_SCOPED;
+    llassert(!on_main_thread());
+
+    {
+        LL_PROFILE_ZONE_NAMED("cglt - sync");
+        if (gGLManager.mHasSync)
+        {
+            // post a sync to the main thread (will execute before tex name swap lambda below)
+            // glFlush calls here are partly superstitious and partly backed by observation
+            // on AMD hardware
+            glFlush();
+            auto sync = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
+            glFlush();
+            LL::WorkQueue::postMaybe(
+                mMainQueue,
+                [=]()
+                {
+                    LL_PROFILE_ZONE_NAMED("cglt - wait sync");
+                    {
+                        LL_PROFILE_ZONE_NAMED("glWaitSync");
+                        glWaitSync(sync, 0, GL_TIMEOUT_IGNORED);
+                    }
+                    {
+                        LL_PROFILE_ZONE_NAMED("glDeleteSync");
+                        glDeleteSync(sync);
+                    }
+                });
+        }
+        else
+        {
+            glFinish();
+        }
+    }
+
+    ref();
+    LL::WorkQueue::postMaybe(
+        mMainQueue,
+        [=]()
+        {
+            LL_PROFILE_ZONE_NAMED("cglt - delete callback");
+            syncTexName(new_tex_name);
+            unref();
+        });
+}
+
+void LLImageGL::syncTexName(LLGLuint texname)
+{
+    if (texname != 0)
+    {
+        if (mTexName != 0 && mTexName != texname)
+        {
+            LLImageGL::deleteTextures(1, &mTexName);
+        }
+        mTexName = texname;
+    }
 }
 
 BOOL LLImageGL::readBackRaw(S32 discard_level, LLImageRaw* imageraw, bool compressed_ok) const
@@ -1673,37 +1849,29 @@ BOOL LLImageGL::readBackRaw(S32 discard_level, LLImageRaw* imageraw, bool compre
 	return TRUE ;
 }
 
-void LLImageGL::deleteDeadTextures()
-{
-	bool reset = false;
-
-	if (reset)
-	{
-		gGL.getTexUnit(0)->activate();
-	}
-}
-		
 void LLImageGL::destroyGLTexture()
 {
+    checkActiveThread();
+
 	if (mTexName != 0)
 	{
 		if(mTextureMemory != S32Bytes(0))
 		{
 			sGlobalTextureMemory -= mTextureMemory;
-			disclaimMem(mTextureMemory);
 			mTextureMemory = (S32Bytes)0;
 		}
 		
-		LLImageGL::deleteTextures(1, &mTexName);			
+		LLImageGL::deleteTextures(1, &mTexName);
 		mCurrentDiscardLevel = -1 ; //invalidate mCurrentDiscardLevel.
 		mTexName = 0;		
 		mGLTextureCreated = FALSE ;
-	}	
+	}
 }
 
 //force to invalidate the gl texture, most likely a sculpty texture
 void LLImageGL::forceToInvalidateGLTexture()
 {
+    checkActiveThread();
 	if (mTexName != 0)
 	{
 		destroyGLTexture();
@@ -1878,6 +2046,7 @@ void LLImageGL::calcAlphaChannelOffsetAndStride()
     case GL_LUMINANCE_ALPHA:
         mAlphaStride = 2;
         break;
+    case GL_RED:
     case GL_RGB:
     case GL_SRGB:
         mNeedsAlphaAndPickMask = FALSE;
@@ -2043,7 +2212,6 @@ U32 LLImageGL::createPickMask(S32 pWidth, S32 pHeight)
 	U32 size = pick_width * pick_height;
 	size = (size + 7) / 8; // pixelcount-to-bits
 	mPickMask = new U8[size];
-	claimMem(size);
 	mPickMaskWidth = pick_width - 1;
 	mPickMaskHeight = pick_height - 1;
 
@@ -2058,7 +2226,6 @@ void LLImageGL::freePickMask()
 	// pickmask validity depends on old image size, delete it
 	if (mPickMask != NULL)
 	{
-		disclaimMem((mPickMaskWidth * mPickMaskHeight + 7) / 8);
 		delete [] mPickMask;
 	}
 	mPickMask = NULL;
@@ -2182,6 +2349,12 @@ void LLImageGL::resetCurTexSizebar()
 	sCurTexPickSize = -1 ;
 }
 //----------------------------------------------------------------------------
+#if LL_IMAGEGL_THREAD_CHECK
+void LLImageGL::checkActiveThread()
+{
+    llassert(mActiveThread == LLThread::currentID());
+}
+#endif
 
 //----------------------------------------------------------------------------
 
@@ -2235,3 +2408,38 @@ void LLImageGL::resetCurTexSizebar()
 		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, 0);
 		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL,  nummips);
 */  
+
+std::atomic<S32> LLImageGLThread::sFreeVRAMMegabytes(4096); //if free vram is unknown, default to 4GB
+
+LLImageGLThread::LLImageGLThread(LLWindow* window)
+    // We want exactly one thread, but a very large capacity: we never want
+    // anyone, especially inner-loop render code, to have to block on post()
+    // because we're full.
+    : ThreadPool("LLImageGL", 1, 1024*1024)
+    , mWindow(window)
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    sEnabled = true;
+    mFinished = false;
+
+    mContext = mWindow->createSharedContext();
+    ThreadPool::start();
+}
+
+void LLImageGLThread::run()
+{
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_TEXTURE;
+    // We must perform setup on this thread before actually servicing our
+    // WorkQueue, likewise cleanup afterwards.
+    mWindow->makeContextCurrent(mContext);
+    gGL.init();
+    ThreadPool::run();
+    gGL.shutdown();
+    mWindow->destroySharedContext(mContext);
+}
+
+S32 LLImageGLThread::getFreeVRAMMegabytes()
+{
+    return sFreeVRAMMegabytes;
+}
+
diff --git a/indra/llrender/llimagegl.h b/indra/llrender/llimagegl.h
index 61ddc8d59b..4d5b60d6bc 100644
--- a/indra/llrender/llimagegl.h
+++ b/indra/llrender/llimagegl.h
@@ -35,21 +35,26 @@
 #include "llrefcount.h"
 #include "v2math.h"
 #include "llunits.h"
-
+#include "llthreadsafequeue.h"
 #include "llrender.h"
-class LLTextureAtlas ;
+#include "threadpool.h"
+#include "workqueue.h"
+
+#define LL_IMAGEGL_THREAD_CHECK 0 //set to 1 to enable thread debugging for ImageGL
+
+class LLWindow;
+
 #define BYTES_TO_MEGA_BYTES(x) ((x) >> 20)
 #define MEGA_BYTES_TO_BYTES(x) ((x) << 20)
 
 //============================================================================
-class LLImageGL : public LLRefCount, public LLTrace::MemTrackable<LLImageGL>
+class LLImageGL : public LLRefCount
 {
 	friend class LLTexUnit;
 public:
 	// These 2 functions replace glGenTextures() and glDeleteTextures()
 	static void generateTextures(S32 numTextures, U32 *textures);
-	static void deleteTextures(S32 numTextures, U32 *textures);
-	static void deleteDeadTextures();
+	static void deleteTextures(S32 numTextures, const U32 *textures);
 
 	// Size calculation
 	static S32 dataFormatBits(S32 dataformat);
@@ -102,17 +107,21 @@ public:
 	void setAllowCompression(bool allow) { mAllowCompression = allow; }
 
 	static void setManualImage(U32 target, S32 miplevel, S32 intformat, S32 width, S32 height, U32 pixformat, U32 pixtype, const void *pixels, bool allow_compression = true);
-
+    
 	BOOL createGLTexture() ;
 	BOOL createGLTexture(S32 discard_level, const LLImageRaw* imageraw, S32 usename = 0, BOOL to_create = TRUE,
-		S32 category = sMaxCategories-1);
-	BOOL createGLTexture(S32 discard_level, const U8* data, BOOL data_hasmips = FALSE, S32 usename = 0);
+		S32 category = sMaxCategories-1, bool defer_copy = false, LLGLuint* tex_name = nullptr);
+	BOOL createGLTexture(S32 discard_level, const U8* data, BOOL data_hasmips = FALSE, S32 usename = 0, bool defer_copy = false, LLGLuint* tex_name = nullptr);
 	void setImage(const LLImageRaw* imageraw);
-	BOOL setImage(const U8* data_in, BOOL data_hasmips = FALSE);
-	BOOL setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update = FALSE);
-	BOOL setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update = FALSE);
+	BOOL setImage(const U8* data_in, BOOL data_hasmips = FALSE, S32 usename = 0);
+	BOOL setSubImage(const LLImageRaw* imageraw, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update = FALSE, LLGLuint use_name = 0);
+	BOOL setSubImage(const U8* datap, S32 data_width, S32 data_height, S32 x_pos, S32 y_pos, S32 width, S32 height, BOOL force_fast_update = FALSE, LLGLuint use_name = 0);
 	BOOL setSubImageFromFrameBuffer(S32 fb_x, S32 fb_y, S32 x_pos, S32 y_pos, S32 width, S32 height);
-	
+
+    // wait for gl commands to finish on current thread and push
+    // a lambda to main thread to swap mNewTexName and mTexName
+    void syncToMainThread(LLGLuint new_tex_name);
+
 	// Read back a raw image for this discard level, if it exists
 	BOOL readBackRaw(S32 discard_level, LLImageRaw* imageraw, bool compressed_ok) const;
 	void destroyGLTexture();
@@ -184,6 +193,12 @@ public:
 	BOOL preAddToAtlas(S32 discard_level, const LLImageRaw* raw_image);
 	void postAddToAtlas() ;	
 
+#if LL_IMAGEGL_THREAD_CHECK
+    // thread debugging
+    std::thread::id mActiveThread;
+    void checkActiveThread();
+#endif
+
 public:
 	// Various GL/Rendering options
 	S32Bytes mTextureMemory;
@@ -194,6 +209,7 @@ private:
 	void freePickMask();
 
 	LLPointer<LLImageRaw> mSaveData; // used for destroyGL/restoreGL
+	LL::WorkQueue::weak_t mMainQueue;
 	U8* mPickMask;  //downsampled bitmap approximation of alpha channel.  NULL if no alpha channel
 	U16 mPickMaskWidth;
 	U16 mPickMaskHeight;
@@ -208,8 +224,9 @@ private:
 
 	bool     mGLTextureCreated ;
 	LLGLuint mTexName;
+    //LLGLuint mNewTexName = 0; // tex name set by background thread to be applied in main thread
 	U16      mWidth;
-	U16      mHeight;	
+	U16      mHeight;
 	S8       mCurrentDiscardLevel;
 	
 	S8       mDiscardLevelInAtlas;
@@ -265,7 +282,7 @@ public:
 #endif
 
 public:
-	static void initClass(S32 num_catagories, BOOL skip_analyze_alpha = false); 
+	static void initClass(LLWindow* window, S32 num_catagories, BOOL skip_analyze_alpha = false, bool multi_threaded = false); 
 	static void cleanupClass() ;
 
 private:
@@ -287,6 +304,9 @@ public:
 	
     void setTexName(GLuint texName) { mTexName = texName; }
 
+    //similar to setTexName, but will call deleteTextures on mTexName if mTexName is not 0 or texname
+    void syncTexName(LLGLuint texname);
+
 	//for debug use: show texture size distribution 
 	//----------------------------------------
 	static S32 sCurTexSizeBar ;
@@ -301,4 +321,36 @@ public:
 
 };
 
+class LLImageGLThread : public LLSimpleton<LLImageGLThread>, LL::ThreadPool
+{
+public:
+    // follows gSavedSettings "RenderGLMultiThreaded"
+    static bool sEnabled;
+    
+    // app should call this function periodically
+    static void updateClass();
+
+    // free video memory in megabytes
+    static std::atomic<S32> sFreeVRAMMegabytes;
+
+    LLImageGLThread(LLWindow* window);
+
+    // post a function to be executed on the LLImageGL background thread
+    template <typename CALLABLE>
+    bool post(CALLABLE&& func)
+    {
+        return getQueue().postIfOpen(std::forward<CALLABLE>(func));
+    }
+
+    void run() override;
+
+    static S32 getFreeVRAMMegabytes();
+
+private:
+    LLWindow* mWindow;
+    void* mContext = nullptr;
+    LLAtomicBool mFinished;
+};
+
+
 #endif // LL_LLIMAGEGL_H
diff --git a/indra/llrender/llrender.cpp b/indra/llrender/llrender.cpp
index 03b6aac20c..a03a27cf94 100644
--- a/indra/llrender/llrender.cpp
+++ b/indra/llrender/llrender.cpp
@@ -36,7 +36,18 @@
 #include "lltexture.h"
 #include "llshadermgr.h"
 
-LLRender gGL;
+#if LL_WINDOWS
+extern void APIENTRY gl_debug_callback(GLenum source,
+                                GLenum type,
+                                GLuint id,
+                                GLenum severity,
+                                GLsizei length,
+                                const GLchar* message,
+                                GLvoid* userParam)
+;
+#endif
+
+thread_local LLRender gGL;
 
 // Handy copies of last good GL matrices
 F32	gGLModelView[16];
@@ -71,18 +82,6 @@ static const GLint sGLAddressMode[] =
 	GL_CLAMP_TO_EDGE
 };
 
-static const GLenum sGLCompareFunc[] =
-{
-	GL_NEVER,
-	GL_ALWAYS,
-	GL_LESS,
-	GL_LEQUAL,
-	GL_EQUAL,
-	GL_NOTEQUAL,
-	GL_GEQUAL,
-	GL_GREATER
-};
-
 const U32 immediate_mask = LLVertexBuffer::MAP_VERTEX | LLVertexBuffer::MAP_COLOR | LLVertexBuffer::MAP_TEXCOORD0;
 
 static const GLenum sGLBlendFactor[] =
@@ -102,10 +101,7 @@ static const GLenum sGLBlendFactor[] =
 };
 
 LLTexUnit::LLTexUnit(S32 index)
-	: mCurrTexType(TT_NONE), mCurrBlendType(TB_MULT), 
-	mCurrColorOp(TBO_MULT), mCurrAlphaOp(TBO_MULT),
-	mCurrColorSrc1(TBS_TEX_COLOR), mCurrColorSrc2(TBS_PREV_COLOR),
-	mCurrAlphaSrc1(TBS_TEX_ALPHA), mCurrAlphaSrc2(TBS_PREV_ALPHA),
+	: mCurrTexType(TT_NONE),
     mCurrColorScale(1), mCurrAlphaScale(1), mCurrTexture(0), mTexColorSpace(TCS_LINEAR),
 	mHasMipMaps(false),
 	mIndex(index)
@@ -128,42 +124,15 @@ void LLTexUnit::refreshState(void)
 	
 	glActiveTextureARB(GL_TEXTURE0_ARB + mIndex);
 
-	//
-	// Per apple spec, don't call glEnable/glDisable when index exceeds max texture units
-	// http://www.mailinglistarchive.com/html/mac-opengl@lists.apple.com/2008-07/msg00653.html
-	//
-	bool enableDisable = !LLGLSLShader::sNoFixedFunction && 
-		(mIndex < gGLManager.mNumTextureUnits) && mCurrTexType != LLTexUnit::TT_MULTISAMPLE_TEXTURE;
-		
 	if (mCurrTexType != TT_NONE)
 	{
-		if (enableDisable)
-		{
-			glEnable(sGLTextureType[mCurrTexType]);
-		}
-		
 		glBindTexture(sGLTextureType[mCurrTexType], mCurrTexture);
 	}
 	else
 	{
-		if (enableDisable)
-		{
-			glDisable(GL_TEXTURE_2D);
-		}
-		
 		glBindTexture(GL_TEXTURE_2D, 0);	
 	}
 
-	if (mCurrBlendType != TB_COMBINE)
-	{
-		setTextureBlendType(mCurrBlendType);
-	}
-	else
-	{
-		setTextureCombiner(mCurrColorOp, mCurrColorSrc1, mCurrColorSrc2, false);
-		setTextureCombiner(mCurrAlphaOp, mCurrAlphaSrc1, mCurrAlphaSrc2, true);
-	}
-
     setTextureColorSpace(mTexColorSpace);
 }
 
@@ -196,14 +165,6 @@ void LLTexUnit::enable(eTextureType type)
 		mCurrTexType = type;
 
 		gGL.flush();
-		if (!LLGLSLShader::sNoFixedFunction && 
-			type != LLTexUnit::TT_MULTISAMPLE_TEXTURE &&
-			mIndex < gGLManager.mNumTextureUnits)
-		{
-			stop_glerror();
-			glEnable(sGLTextureType[type]);
-			stop_glerror();
-		}
 	}
 }
 
@@ -216,21 +177,34 @@ void LLTexUnit::disable(void)
 		activate();
 		unbind(mCurrTexType);
 		gGL.flush();
-		if (!LLGLSLShader::sNoFixedFunction &&
-			mCurrTexType != LLTexUnit::TT_MULTISAMPLE_TEXTURE &&
-			mIndex < gGLManager.mNumTextureUnits)
-		{
-			glDisable(sGLTextureType[mCurrTexType]);
-		}
-
         setTextureColorSpace(TCS_LINEAR);
 		
 		mCurrTexType = TT_NONE;
 	}
 }
 
+void LLTexUnit::bindFast(LLTexture* texture)
+{
+    LLImageGL* gl_tex = texture->getGLTexture();
+
+    glActiveTextureARB(GL_TEXTURE0_ARB + mIndex);
+    gGL.mCurrTextureUnitIndex = mIndex;
+    mCurrTexture = gl_tex->getTexName();
+    if (!mCurrTexture)
+    {
+        LL_PROFILE_ZONE_NAMED("MISSING TEXTURE");
+        //if deleted, will re-generate it immediately
+        texture->forceImmediateUpdate();
+        gl_tex->forceUpdateBindStats();
+        texture->bindDefaultImage(mIndex);
+    }
+    glBindTexture(sGLTextureType[gl_tex->getTarget()], mCurrTexture);
+    mHasMipMaps = gl_tex->mHasMipMaps;
+}
+
 bool LLTexUnit::bind(LLTexture* texture, bool for_rendering, bool forceBind)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_PIPELINE;
 	stop_glerror();
 	if (mIndex >= 0)
 	{
@@ -294,18 +268,20 @@ bool LLTexUnit::bind(LLTexture* texture, bool for_rendering, bool forceBind)
 	return true;
 }
 
-bool LLTexUnit::bind(LLImageGL* texture, bool for_rendering, bool forceBind)
+bool LLTexUnit::bind(LLImageGL* texture, bool for_rendering, bool forceBind, S32 usename)
 {
 	stop_glerror();
 	if (mIndex < 0) return false;
 
+    U32 texname = usename ? usename : texture->getTexName();
+
 	if(!texture)
 	{
 		LL_DEBUGS() << "NULL LLTexUnit::bind texture" << LL_ENDL;
 		return false;
 	}
 
-	if(!texture->getTexName())
+	if(!texname)
 	{
 		if(LLImageGL::sDefaultGLTexture && LLImageGL::sDefaultGLTexture->getTexName())
 		{
@@ -315,7 +291,7 @@ bool LLTexUnit::bind(LLImageGL* texture, bool for_rendering, bool forceBind)
 		return false ;
 	}
 
-	if ((mCurrTexture != texture->getTexName()) || forceBind)
+	if ((mCurrTexture != texname) || forceBind)
 	{
 		gGL.flush();
 		stop_glerror();
@@ -323,7 +299,7 @@ bool LLTexUnit::bind(LLImageGL* texture, bool for_rendering, bool forceBind)
 		stop_glerror();
 		enable(texture->getTarget());
 		stop_glerror();
-		mCurrTexture = texture->getTexName();
+		mCurrTexture = texname;
 		glBindTexture(sGLTextureType[texture->getTarget()], mCurrTexture);
 		stop_glerror();
 		texture->updateBindStats(texture->mTextureMemory);		
@@ -447,7 +423,7 @@ void LLTexUnit::unbind(eTextureType type)
 
         // Always make sure our texture color space is reset to linear.  SRGB sampling should be opt-in in the vast majority of cases.  Also prevents color space "popping".
         mTexColorSpace = TCS_LINEAR;
-		if (LLGLSLShader::sNoFixedFunction && type == LLTexUnit::TT_TEXTURE)
+		if (type == LLTexUnit::TT_TEXTURE)
 		{
 			glBindTexture(sGLTextureType[type], sWhiteTexture);
 		}
@@ -459,6 +435,28 @@ void LLTexUnit::unbind(eTextureType type)
 	}
 }
 
+void LLTexUnit::unbindFast(eTextureType type)
+{
+    activate();
+
+    // Disabled caching of binding state.
+    if (mCurrTexType == type)
+    {
+        mCurrTexture = 0;
+
+        // Always make sure our texture color space is reset to linear.  SRGB sampling should be opt-in in the vast majority of cases.  Also prevents color space "popping".
+        mTexColorSpace = TCS_LINEAR;
+        if (type == LLTexUnit::TT_TEXTURE)
+        {
+            glBindTexture(sGLTextureType[type], sWhiteTexture);
+        }
+        else
+        {
+            glBindTexture(sGLTextureType[type], 0);
+        }
+    }
+}
+
 void LLTexUnit::setTextureAddressMode(eTextureAddressMode mode)
 {
 	if (mIndex < 0 || mCurrTexture == 0) return;
@@ -537,55 +535,6 @@ void LLTexUnit::setTextureFilteringOption(LLTexUnit::eTextureFilterOptions optio
 	}
 }
 
-void LLTexUnit::setTextureBlendType(eTextureBlendType type)
-{
-	if (LLGLSLShader::sNoFixedFunction)
-	{ //texture blend type means nothing when using shaders
-		return;
-	}
-
-	if (mIndex < 0) return;
-
-	// Do nothing if it's already correctly set.
-	if (mCurrBlendType == type && !gGL.mDirty)
-	{
-		return;
-	}
-
-	gGL.flush();
-
-	activate();
-	mCurrBlendType = type;
-	S32 scale_amount = 1;
-	switch (type) 
-	{
-		case TB_REPLACE:
-			glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
-			break;
-		case TB_ADD:
-			glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_ADD);
-			break;
-		case TB_MULT:
-			glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE);
-			break;
-		case TB_MULT_X2:
-			glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE);
-			scale_amount = 2;
-			break;
-		case TB_ALPHA_BLEND:
-			glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_DECAL);
-			break;
-		case TB_COMBINE:
-			glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_COMBINE_ARB);
-			break;
-		default:
-			LL_ERRS() << "Unknown Texture Blend Type: " << type << LL_ENDL;
-			break;
-	}
-	setColorScale(scale_amount);
-	setAlphaScale(1);
-}
-
 GLint LLTexUnit::getTextureSource(eTextureBlendSrc src)
 {
 	switch(src)
@@ -662,159 +611,6 @@ GLint LLTexUnit::getTextureSourceType(eTextureBlendSrc src, bool isAlpha)
 	}
 }
 
-void LLTexUnit::setTextureCombiner(eTextureBlendOp op, eTextureBlendSrc src1, eTextureBlendSrc src2, bool isAlpha)
-{
-	if (LLGLSLShader::sNoFixedFunction)
-	{ //register combiners do nothing when not using fixed function
-		return;
-	}	
-
-	if (mIndex < 0) return;
-
-	activate();
-	if (mCurrBlendType != TB_COMBINE || gGL.mDirty)
-	{
-		mCurrBlendType = TB_COMBINE;
-		gGL.flush();
-		glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_COMBINE_ARB);
-	}
-	
-	// We want an early out, because this function does a LOT of stuff.
-	if ( ( (isAlpha && (mCurrAlphaOp == op) && (mCurrAlphaSrc1 == src1) && (mCurrAlphaSrc2 == src2))
-			|| (!isAlpha && (mCurrColorOp == op) && (mCurrColorSrc1 == src1) && (mCurrColorSrc2 == src2)) ) && !gGL.mDirty)
-	{
-		return;
-	}
-
-	gGL.flush();
-
-	// Get the gl source enums according to the eTextureBlendSrc sources passed in
-	GLint source1 = getTextureSource(src1);
-	GLint source2 = getTextureSource(src2);
-	// Get the gl operand enums according to the eTextureBlendSrc sources passed in
-	GLint operand1 = getTextureSourceType(src1, isAlpha);
-	GLint operand2 = getTextureSourceType(src2, isAlpha);
-	// Default the scale amount to 1
-	S32 scale_amount = 1;
-	GLenum comb_enum, src0_enum, src1_enum, src2_enum, operand0_enum, operand1_enum, operand2_enum;
-	
-	if (isAlpha)
-	{
-		// Set enums to ALPHA ones
-		comb_enum = GL_COMBINE_ALPHA_ARB;
-		src0_enum = GL_SOURCE0_ALPHA_ARB;
-		src1_enum = GL_SOURCE1_ALPHA_ARB;
-		src2_enum = GL_SOURCE2_ALPHA_ARB;
-		operand0_enum = GL_OPERAND0_ALPHA_ARB;
-		operand1_enum = GL_OPERAND1_ALPHA_ARB;
-		operand2_enum = GL_OPERAND2_ALPHA_ARB;
-
-		// cache current combiner
-		mCurrAlphaOp = op;
-		mCurrAlphaSrc1 = src1;
-		mCurrAlphaSrc2 = src2;
-	}
-	else 
-	{
-		// Set enums to RGB ones
-		comb_enum = GL_COMBINE_RGB_ARB;
-		src0_enum = GL_SOURCE0_RGB_ARB;
-		src1_enum = GL_SOURCE1_RGB_ARB;
-		src2_enum = GL_SOURCE2_RGB_ARB;
-		operand0_enum = GL_OPERAND0_RGB_ARB;
-		operand1_enum = GL_OPERAND1_RGB_ARB;
-		operand2_enum = GL_OPERAND2_RGB_ARB;
-
-		// cache current combiner
-		mCurrColorOp = op;
-		mCurrColorSrc1 = src1;
-		mCurrColorSrc2 = src2;
-	}
-
-	switch(op)
-	{
-		case TBO_REPLACE:
-			// Slightly special syntax (no second sources), just set all and return.
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_REPLACE);
-			glTexEnvi(GL_TEXTURE_ENV, src0_enum, source1);
-			glTexEnvi(GL_TEXTURE_ENV, operand0_enum, operand1);
-			(isAlpha) ? setAlphaScale(1) : setColorScale(1);
-			return;
-
-		case TBO_MULT:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_MODULATE);
-			break;
-
-		case TBO_MULT_X2:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_MODULATE);
-			scale_amount = 2;
-			break;
-
-		case TBO_MULT_X4:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_MODULATE);
-			scale_amount = 4;
-			break;
-
-		case TBO_ADD:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_ADD);
-			break;
-
-		case TBO_ADD_SIGNED:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_ADD_SIGNED_ARB);
-			break;
-
-		case TBO_SUBTRACT:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_SUBTRACT_ARB);
-			break;
-
-		case TBO_LERP_VERT_ALPHA:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_INTERPOLATE);
-			glTexEnvi(GL_TEXTURE_ENV, src2_enum, GL_PRIMARY_COLOR_ARB);
-			glTexEnvi(GL_TEXTURE_ENV, operand2_enum, GL_SRC_ALPHA);
-			break;
-
-		case TBO_LERP_TEX_ALPHA:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_INTERPOLATE);
-			glTexEnvi(GL_TEXTURE_ENV, src2_enum, GL_TEXTURE);
-			glTexEnvi(GL_TEXTURE_ENV, operand2_enum, GL_SRC_ALPHA);
-			break;
-
-		case TBO_LERP_PREV_ALPHA:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_INTERPOLATE);
-			glTexEnvi(GL_TEXTURE_ENV, src2_enum, GL_PREVIOUS_ARB);
-			glTexEnvi(GL_TEXTURE_ENV, operand2_enum, GL_SRC_ALPHA);
-			break;
-
-		case TBO_LERP_CONST_ALPHA:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_INTERPOLATE);
-			glTexEnvi(GL_TEXTURE_ENV, src2_enum, GL_CONSTANT_ARB);
-			glTexEnvi(GL_TEXTURE_ENV, operand2_enum, GL_SRC_ALPHA);
-			break;
-
-		case TBO_LERP_VERT_COLOR:
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_INTERPOLATE);
-			glTexEnvi(GL_TEXTURE_ENV, src2_enum, GL_PRIMARY_COLOR_ARB);
-			glTexEnvi(GL_TEXTURE_ENV, operand2_enum, (isAlpha) ? GL_SRC_ALPHA : GL_SRC_COLOR);
-			break;
-
-		default:
-			LL_WARNS() << "Unknown eTextureBlendOp: " << op << ".  Setting op to replace." << LL_ENDL;
-			// Slightly special syntax (no second sources), just set all and return.
-			glTexEnvi(GL_TEXTURE_ENV, comb_enum, GL_REPLACE);
-			glTexEnvi(GL_TEXTURE_ENV, src0_enum, source1);
-			glTexEnvi(GL_TEXTURE_ENV, operand0_enum, operand1);
-			(isAlpha) ? setAlphaScale(1) : setColorScale(1);
-			return;
-	}
-
-	// Set sources, operands, and scale accordingly
-	glTexEnvi(GL_TEXTURE_ENV, src0_enum, source1);
-	glTexEnvi(GL_TEXTURE_ENV, operand0_enum, operand1);
-	glTexEnvi(GL_TEXTURE_ENV, src1_enum, source2);
-	glTexEnvi(GL_TEXTURE_ENV, operand1_enum, operand2);
-	(isAlpha) ? setAlphaScale(scale_amount) : setColorScale(scale_amount);
-}
-
 void LLTexUnit::setColorScale(S32 scale)
 {
 	if (mCurrColorScale != scale || gGL.mDirty)
@@ -903,26 +699,12 @@ LLLightState::LLLightState(S32 index)
 
 void LLLightState::enable()
 {
-	if (!mEnabled)
-	{
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glEnable(GL_LIGHT0+mIndex);
-		}
-		mEnabled = true;
-	}
+	mEnabled = true;
 }
 
 void LLLightState::disable()
 {
-	if (mEnabled)
-	{
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glDisable(GL_LIGHT0+mIndex);
-		}
-		mEnabled = false;
-	}
+	mEnabled = false;
 }
 
 void LLLightState::setDiffuse(const LLColor4& diffuse)
@@ -931,10 +713,6 @@ void LLLightState::setDiffuse(const LLColor4& diffuse)
 	{
 		++gGL.mLightHash;
 		mDiffuse = diffuse;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightfv(GL_LIGHT0+mIndex, GL_DIFFUSE, mDiffuse.mV);
-		}
 	}
 }
 
@@ -962,10 +740,6 @@ void LLLightState::setAmbient(const LLColor4& ambient)
 	{
 		++gGL.mLightHash;
 		mAmbient = ambient;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightfv(GL_LIGHT0+mIndex, GL_AMBIENT, mAmbient.mV);
-		}
 	}
 }
 
@@ -975,10 +749,6 @@ void LLLightState::setSpecular(const LLColor4& specular)
 	{
 		++gGL.mLightHash;
 		mSpecular = specular;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightfv(GL_LIGHT0+mIndex, GL_SPECULAR, mSpecular.mV);
-		}
 	}
 }
 
@@ -987,20 +757,11 @@ void LLLightState::setPosition(const LLVector4& position)
 	//always set position because modelview matrix may have changed
 	++gGL.mLightHash;
 	mPosition = position;
-	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);
-	}
-
+	//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)
@@ -1009,10 +770,6 @@ void LLLightState::setConstantAttenuation(const F32& atten)
 	{
 		mConstantAtten = atten;
 		++gGL.mLightHash;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightf(GL_LIGHT0+mIndex, GL_CONSTANT_ATTENUATION, atten);
-		}
 	}
 }
 
@@ -1022,10 +779,6 @@ void LLLightState::setLinearAttenuation(const F32& atten)
 	{
 		++gGL.mLightHash;
 		mLinearAtten = atten;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightf(GL_LIGHT0+mIndex, GL_LINEAR_ATTENUATION, atten);
-		}
 	}
 }
 
@@ -1035,10 +788,6 @@ void LLLightState::setQuadraticAttenuation(const F32& atten)
 	{
 		++gGL.mLightHash;
 		mQuadraticAtten = atten;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightf(GL_LIGHT0+mIndex, GL_QUADRATIC_ATTENUATION, atten);
-		}
 	}
 }
 
@@ -1048,10 +797,6 @@ void LLLightState::setSpotExponent(const F32& exponent)
 	{
 		++gGL.mLightHash;
 		mSpotExponent = exponent;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightf(GL_LIGHT0+mIndex, GL_SPOT_EXPONENT, exponent);
-		}
 	}
 }
 
@@ -1061,10 +806,6 @@ void LLLightState::setSpotCutoff(const F32& cutoff)
 	{
 		++gGL.mLightHash;
 		mSpotCutoff = cutoff;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightf(GL_LIGHT0+mIndex, GL_SPOT_CUTOFF, cutoff);
-		}
 	}
 }
 
@@ -1073,19 +814,12 @@ void LLLightState::setSpotDirection(const LLVector3& direction)
 	//always set direction because modelview matrix may have changed
 	++gGL.mLightHash;
 	mSpotDirection = direction;
-	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);
+	//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);
-	}
+	mSpotDirection.set(dir.v);
 }
 
 LLRender::LLRender()
@@ -1113,8 +847,6 @@ LLRender::LLRender()
 		mCurrColorMask[i] = true;
 	}
 
-	mCurrAlphaFunc = CF_DEFAULT;
-	mCurrAlphaFuncVal = 0.01f;
 	mCurrBlendColorSFactor = BF_UNDEF;
 	mCurrBlendAlphaSFactor = BF_UNDEF;
 	mCurrBlendColorDFactor = BF_UNDEF;
@@ -1139,6 +871,23 @@ LLRender::~LLRender()
 
 void LLRender::init()
 {
+#if LL_WINDOWS
+    if (gGLManager.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
+
+    glPixelStorei(GL_PACK_ALIGNMENT, 1);
+    glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
+
+    gGL.setSceneBlendType(LLRender::BT_ALPHA);
+    gGL.setAmbientLightColor(LLColor4::black);
+
+    glCullFace(GL_BACK);
+
 	if (sGLCoreProfile && !LLVertexBuffer::sUseVAO)
 	{ //bind a dummy vertex array object so we're core profile compliant
 #ifdef GL_ARB_vertex_array_object
@@ -1192,7 +941,7 @@ void LLRender::refreshState(void)
 
 	setColorMask(mCurrColorMask[0], mCurrColorMask[1], mCurrColorMask[2], mCurrColorMask[3]);
 	
-	setAlphaRejectSettings(mCurrAlphaFunc, mCurrAlphaFuncVal);
+    flush();
 
 	mDirty = false;
 }
@@ -1243,9 +992,7 @@ void LLRender::syncLightState()
 
 void LLRender::syncMatrices()
 {
-	stop_glerror();
-
-	static const U32 name[] = 
+    static const U32 name[] = 
 	{
 		LLShaderMgr::MODELVIEW_MATRIX,
 		LLShaderMgr::PROJECTION_MATRIX,
@@ -1382,41 +1129,6 @@ void LLRender::syncMatrices()
 			syncLightState();
 		}
 	}
-	else if (!LLGLSLShader::sNoFixedFunction)
-	{
-		static const GLenum mode[] = 
-		{
-			GL_MODELVIEW,
-			GL_PROJECTION,
-			GL_TEXTURE,
-			GL_TEXTURE,
-			GL_TEXTURE,
-			GL_TEXTURE,
-		};
-
-		for (U32 i = 0; i < MM_TEXTURE0; ++i)
-		{
-			if (mMatHash[i] != mCurMatHash[i])
-			{
-				glMatrixMode(mode[i]);
-				glLoadMatrixf(mMatrix[i][mMatIdx[i]].m);
-				mCurMatHash[i] = mMatHash[i];
-			}
-		}
-
-		for (U32 i = MM_TEXTURE0; i < NUM_MATRIX_MODES; ++i)
-		{
-			if (mMatHash[i] != mCurMatHash[i])
-			{
-				gGL.getTexUnit(i-MM_TEXTURE0)->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)
@@ -1732,55 +1444,6 @@ void LLRender::setSceneBlendType(eBlendType type)
 	}
 }
 
-void LLRender::setAlphaRejectSettings(eCompareFunc func, F32 value)
-{
-	flush();
-
-	if (LLGLSLShader::sNoFixedFunction)
-	{ //glAlphaFunc is deprecated in OpenGL 3.3
-		return;
-	}
-
-	if (mCurrAlphaFunc != func ||
-		mCurrAlphaFuncVal != value)
-	{
-		mCurrAlphaFunc = func;
-		mCurrAlphaFuncVal = value;
-		if (func == CF_DEFAULT)
-		{
-			glAlphaFunc(GL_GREATER, 0.01f);
-		} 
-		else
-		{
-			glAlphaFunc(sGLCompareFunc[func], value);
-		}
-	}
-
-	if (gDebugGL)
-	{ //make sure cached state is correct
-		GLint cur_func = 0;
-		glGetIntegerv(GL_ALPHA_TEST_FUNC, &cur_func);
-
-		if (func == CF_DEFAULT)
-		{
-			func = CF_GREATER;
-		}
-
-		if (cur_func != sGLCompareFunc[func])
-		{
-			LL_ERRS() << "Alpha test function corrupted!" << LL_ENDL;
-		}
-
-		F32 ref = 0.f;
-		glGetFloatv(GL_ALPHA_TEST_REF, &ref);
-
-		if (ref != value)
-		{
-			LL_ERRS() << "Alpha test value corrupted!" << LL_ENDL;
-		}
-	}
-}
-
 void LLRender::blendFunc(eBlendFactor sfactor, eBlendFactor dfactor)
 {
 	llassert(sfactor < BF_UNDEF);
@@ -1848,14 +1511,11 @@ LLLightState* LLRender::getLight(U32 index)
 
 void LLRender::setAmbientLightColor(const LLColor4& color)
 {
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_PIPELINE
 	if (color != mAmbientLightColor)
 	{
 		++mLightHash;
 		mAmbientLightColor = color;
-		if (!LLGLSLShader::sNoFixedFunction)
-		{
-			glLightModelfv(GL_LIGHT_MODEL_AMBIENT, color.mV);
-		}
 	}
 }
 
@@ -1926,6 +1586,7 @@ void LLRender::flush()
 {
 	if (mCount > 0)
 	{
+        LL_PROFILE_ZONE_SCOPED_CATEGORY_PIPELINE;
 		if (!mUIOffset.empty())
 		{
 			sUICalls++;
@@ -2307,7 +1968,7 @@ void LLRender::color3fv(const GLfloat* c)
 void LLRender::diffuseColor3f(F32 r, F32 g, F32 b)
 {
 	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
-	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+	llassert(shader != NULL);
 
 	if (shader)
 	{
@@ -2322,7 +1983,7 @@ void LLRender::diffuseColor3f(F32 r, F32 g, F32 b)
 void LLRender::diffuseColor3fv(const F32* c)
 {
 	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
-	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+	llassert(shader != NULL);
 
 	if (shader)
 	{
@@ -2337,7 +1998,7 @@ void LLRender::diffuseColor3fv(const F32* c)
 void LLRender::diffuseColor4f(F32 r, F32 g, F32 b, F32 a)
 {
 	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
-	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+	llassert(shader != NULL);
 
 	if (shader)
 	{
@@ -2352,7 +2013,7 @@ void LLRender::diffuseColor4f(F32 r, F32 g, F32 b, F32 a)
 void LLRender::diffuseColor4fv(const F32* c)
 {
 	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
-	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+	llassert(shader != NULL);
 
 	if (shader)
 	{
@@ -2367,7 +2028,7 @@ void LLRender::diffuseColor4fv(const F32* c)
 void LLRender::diffuseColor4ubv(const U8* c)
 {
 	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
-	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+	llassert(shader != NULL);
 
 	if (shader)
 	{
@@ -2382,7 +2043,7 @@ void LLRender::diffuseColor4ubv(const U8* c)
 void LLRender::diffuseColor4ub(U8 r, U8 g, U8 b, U8 a)
 {
 	LLGLSLShader* shader = LLGLSLShader::sCurBoundShaderPtr;
-	llassert(!LLGLSLShader::sNoFixedFunction || shader != NULL);
+	llassert(shader != NULL);
 
 	if (shader)
 	{
diff --git a/indra/llrender/llrender.h b/indra/llrender/llrender.h
index af8568f8a3..e2489876e4 100644
--- a/indra/llrender/llrender.h
+++ b/indra/llrender/llrender.h
@@ -105,10 +105,7 @@ public:
 		TBO_LERP_VERT_ALPHA,		// Interpolate based on Vertex Alpha (VA): ( Source1 * VA + Source2 * (1-VA) )
 		TBO_LERP_TEX_ALPHA,			// Interpolate based on Texture Alpha (TA): ( Source1 * TA + Source2 * (1-TA) )
 		TBO_LERP_PREV_ALPHA,		// Interpolate based on Previous Alpha (PA): ( Source1 * PA + Source2 * (1-PA) )
-		TBO_LERP_CONST_ALPHA,		// Interpolate based on Const Alpha (CA): ( Source1 * CA + Source2 * (1-CA) )
-		TBO_LERP_VERT_COLOR			// Interpolate based on Vertex Col (VC): ( Source1 * VC + Source2 * (1-VC) )
-										// *Note* TBO_LERP_VERTEX_COLOR only works with setTextureColorBlend(),
-										// and falls back to TBO_LERP_VERTEX_ALPHA for setTextureAlphaBlend().
+		TBO_LERP_CONST_ALPHA		// Interpolate based on Const Alpha (CA): ( Source1 * CA + Source2 * (1-CA) )
 	} eTextureBlendOp;
 
 	typedef enum 
@@ -158,9 +155,20 @@ public:
 	
 	// Binds the LLImageGL to this texture unit 
 	// (automatically enables the unit for the LLImageGL's texture type)
-	bool bind(LLImageGL* texture, bool for_rendering = false, bool forceBind = false);
+	bool bind(LLImageGL* texture, bool for_rendering = false, bool forceBind = false, S32 usename = 0);
     bool bind(LLTexture* texture, bool for_rendering = false, bool forceBind = false);
 
+    // bind implementation for inner loops
+    // makes the following assumptions:
+    //  - No need for gGL.flush() 
+    //  - texture is not null
+    //  - gl_tex->getTexName() is not zero
+    //  - This texture is not being bound redundantly
+    //  - USE_SRGB_DECODE is disabled
+    //  - mTexOptionsDirty is false
+    //  - 
+    void bindFast(LLTexture* texture);
+
 	// Binds a cubemap to this texture unit 
 	// (automatically enables the texture unit for cubemaps)
 	bool bind(LLCubeMap* cubeMap);
@@ -177,6 +185,9 @@ public:
 	// (only if there's a texture of the given type currently bound)
 	void unbind(eTextureType type);
 
+    // Fast but unsafe version of unbind
+    void unbindFast(eTextureType type);
+
 	// Sets the addressing mode used to sample the texture
 	// Warning: this stays set for the bound texture forever, 
 	// make sure you want to permanently change the address mode  for the bound texture.
@@ -187,15 +198,6 @@ public:
 	// make sure you want to permanently change the filtering for the bound texture.
 	void setTextureFilteringOption(LLTexUnit::eTextureFilterOptions option);
 
-	void setTextureBlendType(eTextureBlendType type);
-
-	inline void setTextureColorBlend(eTextureBlendOp op, eTextureBlendSrc src1, eTextureBlendSrc src2 = TBS_PREV_COLOR)
-	{ setTextureCombiner(op, src1, src2, false); }
-
-	// NOTE: If *_COLOR enums are passed to src1 or src2, the corresponding *_ALPHA enum will be used instead.
-	inline void setTextureAlphaBlend(eTextureBlendOp op, eTextureBlendSrc src1, eTextureBlendSrc src2 = TBS_PREV_ALPHA)
-	{ setTextureCombiner(op, src1, src2, true); }
-
 	static U32 getInternalType(eTextureType type);
 
 	U32 getCurrTexture(void) { return mCurrTexture; }
@@ -212,13 +214,6 @@ protected:
 	const S32			mIndex;
 	U32					mCurrTexture;
 	eTextureType		mCurrTexType;
-	eTextureBlendType	mCurrBlendType;
-	eTextureBlendOp		mCurrColorOp;
-	eTextureBlendSrc	mCurrColorSrc1;
-	eTextureBlendSrc	mCurrColorSrc2;
-	eTextureBlendOp		mCurrAlphaOp;
-	eTextureBlendSrc	mCurrAlphaSrc1;
-	eTextureBlendSrc	mCurrAlphaSrc2;
     eTextureColorSpace  mTexColorSpace;
 	S32					mCurrColorScale;
 	S32					mCurrAlphaScale;
@@ -229,7 +224,6 @@ protected:
 	void setAlphaScale(S32 scale);
 	GLint getTextureSource(eTextureBlendSrc src);
 	GLint getTextureSourceType(eTextureBlendSrc src, bool isAlpha = false);
-	void setTextureCombiner(eTextureBlendOp op, eTextureBlendSrc src1, eTextureBlendSrc src2, bool isAlpha = false);
 };
 
 class LLLightState
@@ -430,8 +424,6 @@ public:
 	void setColorMask(bool writeColorR, bool writeColorG, bool writeColorB, bool writeAlpha);
 	void setSceneBlendType(eBlendType type);
 
-	void setAlphaRejectSettings(eCompareFunc func, F32 value = 0.01f);
-
 	// applies blend func to both color and alpha
 	void blendFunc(eBlendFactor sfactor, eBlendFactor dfactor);
 	// applies separate blend functions to color and alpha
@@ -482,8 +474,6 @@ private:
 	U32				mMode;
 	U32				mCurrTextureUnitIndex;
 	bool				mCurrColorMask[4];
-	eCompareFunc			mCurrAlphaFunc;
-	F32				mCurrAlphaFuncVal;
 
 	LLPointer<LLVertexBuffer>	mBuffer;
 	LLStrider<LLVector3>		mVerticesp;
@@ -511,7 +501,7 @@ extern F32 gGLLastProjection[16];
 extern F32 gGLProjection[16];
 extern S32 gGLViewport[4];
 
-extern LLRender gGL;
+extern thread_local LLRender gGL;
 
 // This rotation matrix moves the default OpenGL reference frame 
 // (-Z at, Y up) to Cory's favorite reference frame (X at, Z up)
diff --git a/indra/llrender/llrender2dutils.cpp b/indra/llrender/llrender2dutils.cpp
index dd34f3e383..5cb1dc6b25 100644
--- a/indra/llrender/llrender2dutils.cpp
+++ b/indra/llrender/llrender2dutils.cpp
@@ -132,40 +132,15 @@ void gl_rect_2d(S32 left, S32 top, S32 right, S32 bottom, BOOL filled )
 	}
 	else
 	{
-		if( gGLManager.mATIOffsetVerticalLines )
-		{
-			// Work around bug in ATI driver: vertical lines are offset by (-1,-1)
-			gGL.begin( LLRender::LINES );
-
-				// Verticals 
-				gGL.vertex2i(left + 1, top);
-				gGL.vertex2i(left + 1, bottom);
-
-				gGL.vertex2i(right, bottom);
-				gGL.vertex2i(right, top);
-
-				// Horizontals
-				top--;
-				right--;
-				gGL.vertex2i(left, bottom);
-				gGL.vertex2i(right, bottom);
-
-				gGL.vertex2i(left, top);
-				gGL.vertex2i(right, top);
-			gGL.end();
-		}
-		else
-		{
-			top--;
-			right--;
-			gGL.begin( LLRender::LINE_STRIP );
-				gGL.vertex2i(left, top);
-				gGL.vertex2i(left, bottom);
-				gGL.vertex2i(right, bottom);
-				gGL.vertex2i(right, top);
-				gGL.vertex2i(left, top);
-			gGL.end();
-		}
+		top--;
+		right--;
+		gGL.begin( LLRender::LINE_STRIP );
+			gGL.vertex2i(left, top);
+			gGL.vertex2i(left, bottom);
+			gGL.vertex2i(right, bottom);
+			gGL.vertex2i(right, top);
+			gGL.vertex2i(left, top);
+		gGL.end();
 	}
 	stop_glerror();
 }
@@ -250,15 +225,6 @@ void gl_drop_shadow(S32 left, S32 top, S32 right, S32 bottom, const LLColor4 &st
 
 void gl_line_2d(S32 x1, S32 y1, S32 x2, S32 y2 )
 {
-	// Work around bug in ATI driver: vertical lines are offset by (-1,-1)
-	if( (x1 == x2) && gGLManager.mATIOffsetVerticalLines )
-	{
-		x1++;
-		x2++;
-		y1++;
-		y2++;
-	}
-
 	gGL.getTexUnit(0)->unbind(LLTexUnit::TT_TEXTURE);
 	
 	gGL.begin(LLRender::LINES);
@@ -269,15 +235,6 @@ void gl_line_2d(S32 x1, S32 y1, S32 x2, S32 y2 )
 
 void gl_line_2d(S32 x1, S32 y1, S32 x2, S32 y2, const LLColor4 &color )
 {
-	// Work around bug in ATI driver: vertical lines are offset by (-1,-1)
-	if( (x1 == x2) && gGLManager.mATIOffsetVerticalLines )
-	{
-		x1++;
-		x2++;
-		y1++;
-		y2++;
-	}
-
 	gGL.getTexUnit(0)->unbind(LLTexUnit::TT_TEXTURE);
 
 	gGL.color4fv( color.mV );
@@ -395,15 +352,7 @@ void gl_draw_scaled_image_with_border(S32 x, S32 y, S32 width, S32 height, LLTex
 
 	if (solid_color)
 	{
-		if (LLGLSLShader::sNoFixedFunction)
-		{
-			gSolidColorProgram.bind();
-		}
-		else
-		{
-			gGL.getTexUnit(0)->setTextureColorBlend(LLTexUnit::TBO_REPLACE, LLTexUnit::TBS_PREV_COLOR);
-			gGL.getTexUnit(0)->setTextureAlphaBlend(LLTexUnit::TBO_MULT, LLTexUnit::TBS_TEX_ALPHA, LLTexUnit::TBS_VERT_ALPHA);
-		}
+		gSolidColorProgram.bind();
 	}
 
 	if (center_rect.mLeft == 0.f
@@ -650,14 +599,7 @@ void gl_draw_scaled_image_with_border(S32 x, S32 y, S32 width, S32 height, LLTex
 
 	if (solid_color)
 	{
-		if (LLGLSLShader::sNoFixedFunction)
-		{
-			gUIProgram.bind();
-		}
-		else
-		{
-			gGL.getTexUnit(0)->setTextureBlendType(LLTexUnit::TB_MULT);
-		}
+		gUIProgram.bind();
 	}
 }
 
@@ -774,10 +716,6 @@ void gl_draw_scaled_rotated_image(S32 x, S32 y, S32 width, S32 height, F32 degre
 
 void gl_stippled_line_3d( const LLVector3& start, const LLVector3& end, const LLColor4& color, F32 phase )
 {
-	phase = fmod(phase, 1.f);
-
-	S32 shift = S32(phase * 4.f) % 4;
-
 	// Stippled line
 	LLGLEnable stipple(GL_LINE_STIPPLE);
 	
@@ -786,11 +724,6 @@ void gl_stippled_line_3d( const LLVector3& start, const LLVector3& end, const LL
 	gGL.flush();
 	glLineWidth(2.5f);
 
-	if (!LLGLSLShader::sNoFixedFunction)
-	{
-		glLineStipple(2, 0x3333 << shift);
-	}
-
 	gGL.begin(LLRender::LINES);
 	{
 		gGL.vertex3fv( start.mV );
@@ -930,52 +863,16 @@ void gl_ring( F32 radius, F32 width, const LLColor4& center_color, const LLColor
 // Draw gray and white checkerboard with black border
 void gl_rect_2d_checkerboard(const LLRect& rect, GLfloat alpha)
 {
-	if (!LLGLSLShader::sNoFixedFunction)
-	{ 
-	// Initialize the first time this is called.
-	const S32 PIXELS = 32;
-	static GLubyte checkerboard[PIXELS * PIXELS];
-	static BOOL first = TRUE;
-	if( first )
-	{
-		for( S32 i = 0; i < PIXELS; i++ )
-		{
-			for( S32 j = 0; j < PIXELS; j++ )
-			{
-				checkerboard[i * PIXELS + j] = ((i & 1) ^ (j & 1)) * 0xFF;
-			}
-		}
-		first = FALSE;
-	}
-	
-	gGL.getTexUnit(0)->unbind(LLTexUnit::TT_TEXTURE);
-
-	// ...white squares
-	gGL.color4f( 1.f, 1.f, 1.f, alpha );
-	gl_rect_2d(rect);
-
-	// ...gray squares
-	gGL.color4f( .7f, .7f, .7f, alpha );
-	gGL.flush();
+	//polygon stipple is deprecated, use "Checker" texture
+	LLPointer<LLUIImage> img = LLRender2D::getInstance()->getUIImage("Checker");
+	gGL.getTexUnit(0)->bind(img->getImage());
+	gGL.getTexUnit(0)->setTextureAddressMode(LLTexUnit::TAM_WRAP);
+	gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_POINT);
 
-		glPolygonStipple( checkerboard );
+	LLColor4 color(1.f, 1.f, 1.f, alpha);
+	LLRectf uv_rect(0, 0, rect.getWidth()/32.f, rect.getHeight()/32.f);
 
-		LLGLEnable polygon_stipple(GL_POLYGON_STIPPLE);
-		gl_rect_2d(rect);
-	}
-	else
-	{ //polygon stipple is deprecated, use "Checker" texture
-		LLPointer<LLUIImage> img = LLRender2D::getInstance()->getUIImage("Checker");
-		gGL.getTexUnit(0)->bind(img->getImage());
-		gGL.getTexUnit(0)->setTextureAddressMode(LLTexUnit::TAM_WRAP);
-		gGL.getTexUnit(0)->setTextureFilteringOption(LLTexUnit::TFO_POINT);
-
-		LLColor4 color(1.f, 1.f, 1.f, alpha);
-		LLRectf uv_rect(0, 0, rect.getWidth()/32.f, rect.getHeight()/32.f);
-
-		gl_draw_scaled_image(rect.mLeft, rect.mBottom, rect.getWidth(), rect.getHeight(),
-			img->getImage(), color, uv_rect);
-	}
+	gl_draw_scaled_image(rect.mLeft, rect.mBottom, rect.getWidth(), rect.getHeight(), img->getImage(), color, uv_rect);
 	
 	gGL.flush();
 }
@@ -1083,8 +980,6 @@ void gl_rect_2d_simple( S32 width, S32 height )
 	gGL.end();
 }
 
-static LLTrace::BlockTimerStatHandle FTM_RENDER_SEGMENTED_RECT ("Render segmented rectangle");
-
 void gl_segmented_rect_2d_tex(const S32 left, 
 							  const S32 top, 
 							  const S32 right, 
@@ -1094,7 +989,7 @@ void gl_segmented_rect_2d_tex(const S32 left,
 							  const S32 border_size, 
 							  const U32 edges)
 {
-	LL_RECORD_BLOCK_TIME(FTM_RENDER_SEGMENTED_RECT);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_UI;
 
 	S32 width = llabs(right - left);
 	S32 height = llabs(top - bottom);
@@ -1253,7 +1148,7 @@ void gl_segmented_rect_2d_fragment_tex(const LLRect& rect,
 	const F32 end_fragment, 
 	const U32 edges)
 {
-	LL_RECORD_BLOCK_TIME(FTM_RENDER_SEGMENTED_RECT);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_UI;
 	const S32 left = rect.mLeft;
 	const S32 right = rect.mRight;
 	const S32 top = rect.mTop;
@@ -1440,7 +1335,7 @@ void gl_segmented_rect_2d_fragment_tex(const LLRect& rect,
 void gl_segmented_rect_3d_tex(const LLRectf& clip_rect, const LLRectf& center_uv_rect, const LLRectf& center_draw_rect, 
 							 const LLVector3& width_vec, const LLVector3& height_vec)
 {
-	LL_RECORD_BLOCK_TIME(FTM_RENDER_SEGMENTED_RECT);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_UI;
 
 	gGL.begin(LLRender::QUADS);
 	{
diff --git a/indra/llrender/llrendertarget.cpp b/indra/llrender/llrendertarget.cpp
index e3c0255290..0408010513 100644
--- a/indra/llrender/llrendertarget.cpp
+++ b/indra/llrender/llrendertarget.cpp
@@ -170,6 +170,53 @@ bool LLRenderTarget::allocate(U32 resx, U32 resy, U32 color_fmt, bool depth, boo
 	return addColorAttachment(color_fmt);
 }
 
+void LLRenderTarget::setColorAttachment(LLImageGL* img, LLGLuint use_name)
+{
+    LL_PROFILE_ZONE_SCOPED;
+    llassert(img != nullptr); // img must not be null
+    llassert(sUseFBO); // FBO support must be enabled
+    llassert(mDepth == 0); // depth buffers not supported with this mode
+    llassert(mTex.empty()); // mTex must be empty with this mode (binding target should be done via LLImageGL)
+
+    if (mFBO == 0)
+    {
+        glGenFramebuffers(1, (GLuint*)&mFBO);
+    }
+
+    mResX = img->getWidth();
+    mResY = img->getHeight();
+    mUsage = img->getTarget();
+
+    if (use_name == 0)
+    {
+        use_name = img->getTexName();
+    }
+
+    mTex.push_back(use_name);
+
+    glBindFramebuffer(GL_FRAMEBUFFER, mFBO);
+    glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
+            LLTexUnit::getInternalType(mUsage), use_name, 0);
+        stop_glerror();
+
+    check_framebuffer_status();
+
+    glBindFramebuffer(GL_FRAMEBUFFER, sCurFBO);
+}
+
+void LLRenderTarget::releaseColorAttachment()
+{
+    LL_PROFILE_ZONE_SCOPED;
+    llassert(mTex.size() == 1); //cannot use releaseColorAttachment with LLRenderTarget managed color targets
+    llassert(mFBO != 0);  // mFBO must be valid
+    
+    glBindFramebuffer(GL_FRAMEBUFFER, mFBO);
+    glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, LLTexUnit::getInternalType(mUsage), 0, 0);
+    glBindFramebuffer(GL_FRAMEBUFFER, sCurFBO);
+
+    mTex.clear();
+}
+
 bool LLRenderTarget::addColorAttachment(U32 color_fmt)
 {
 	if (color_fmt == 0)
@@ -437,11 +484,13 @@ void LLRenderTarget::bindTarget()
 									GL_COLOR_ATTACHMENT1,
 									GL_COLOR_ATTACHMENT2,
 									GL_COLOR_ATTACHMENT3};
+			LL_PROFILER_GPU_ZONEC( "gl.DrawBuffersARB", 0x4000FF )
 			glDrawBuffersARB(mTex.size(), drawbuffers);
 		}
 			
 		if (mTex.empty())
 		{ //no color buffer to draw to
+			LL_PROFILER_GPU_ZONEC( "gl.DrawBuffer", 0x0000FF )
 			glDrawBuffer(GL_NONE);
 			glReadBuffer(GL_NONE);
 		}
diff --git a/indra/llrender/llrendertarget.h b/indra/llrender/llrendertarget.h
index 6c07ac5b1c..584f224dca 100644
--- a/indra/llrender/llrendertarget.h
+++ b/indra/llrender/llrendertarget.h
@@ -81,6 +81,24 @@ public:
 	// DO use for render targets that resize often and aren't likely to ruin someone's day if they break
 	void resize(U32 resx, U32 resy);
 
+    //point this render target at a particular LLImageGL
+    //   Intended usage:
+    //      LLRenderTarget target;
+    //      target.addColorAttachment(image);
+    //      target.bindTarget();
+    //      < issue GL calls>
+    //      target.flush();
+    //      target.releaseColorAttachment();
+    // 
+    // attachment -- LLImageGL to render into
+    // use_name -- optional texture name to target instead of attachment->getTexName()
+    // NOTE: setColorAttachment and releaseColorAttachment cannot be used in conjuction with
+    // addColorAttachment, allocateDepth, resize, etc.
+    void setColorAttachment(LLImageGL* attachment, LLGLuint use_name = 0);
+
+    // detach from current color attachment
+    void releaseColorAttachment();
+
 	//add color buffer attachment
 	//limit of 4 color attachments per render target
 	bool addColorAttachment(U32 color_fmt);
diff --git a/indra/llrender/llshadermgr.cpp b/indra/llrender/llshadermgr.cpp
index e8c6295930..c100c182dd 100644
--- a/indra/llrender/llshadermgr.cpp
+++ b/indra/llrender/llshadermgr.cpp
@@ -165,6 +165,7 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 
 	if (features->hasObjectSkinning)
 	{
+        shader->mRiggedVariant = shader;
         if (!shader->attachVertexObject("avatar/objectSkinV.glsl"))
 		{
 			return FALSE;
@@ -599,7 +600,7 @@ void LLShaderMgr::dumpObjectLog(GLhandleARB ret, BOOL warns, const std::string&
 	}
  }
 
-GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shader_level, GLenum type, boost::unordered_map<std::string, std::string>* defines, S32 texture_index_channels)
+GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shader_level, GLenum type, std::unordered_map<std::string, std::string>* defines, S32 texture_index_channels)
 {
 
 // endsure work-around for missing GLSL funcs gets propogated to feature shader files (e.g. srgbF.glsl)
@@ -774,14 +775,14 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 	
 	if (defines)
 	{
-		for (boost::unordered_map<std::string,std::string>::iterator iter = defines->begin(); iter != defines->end(); ++iter)
+		for (std::unordered_map<std::string,std::string>::iterator iter = defines->begin(); iter != defines->end(); ++iter)
 		{
 			std::string define = "#define " + iter->first + " " + iter->second + "\n";
 			extra_code_text[extra_code_count++] = (GLcharARB *) strdup(define.c_str());
 		}
 	}
 
-	if( gGLManager.mIsATI )
+	if( gGLManager.mIsAMD )
 	{
 		extra_code_text[extra_code_count++] = strdup( "#define IS_AMD_CARD 1\n" );
 	}
@@ -1053,43 +1054,6 @@ BOOL LLShaderMgr::linkProgramObject(GLhandleARB obj, BOOL suppress_errors)
 		LL_SHADER_LOADING_WARNS() << "GLSL Linker Error:" << LL_ENDL;
 	}
 
-#if LL_DARWIN
-
-	// For some reason this absolutely kills the frame rate when VBO's are enabled
-	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
-		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_SHADER_LOADING_WARNS() << "GLSL Linker: Running in Software:" << LL_ENDL;
-			success = GL_FALSE;
-			suppress_errors = FALSE;		
-		}
-	}
-
-#else
 	std::string log = get_object_log(obj);
 	LLStringUtil::toLower(log);
 	if (log.find("software") != std::string::npos)
@@ -1098,7 +1062,6 @@ BOOL LLShaderMgr::linkProgramObject(GLhandleARB obj, BOOL suppress_errors)
 		success = GL_FALSE;
 		suppress_errors = FALSE;
 	}
-#endif
 	return success;
 }
 
diff --git a/indra/llrender/llshadermgr.h b/indra/llrender/llshadermgr.h
index 3908efd4ec..67c0d6ab10 100644
--- a/indra/llrender/llshadermgr.h
+++ b/indra/llrender/llshadermgr.h
@@ -264,7 +264,7 @@ public:
     void dumpShaderSource(U32 shader_code_count, GLcharARB** shader_code_text);
 	BOOL	linkProgramObject(GLhandleARB obj, BOOL suppress_errors = FALSE);
 	BOOL	validateProgramObject(GLhandleARB obj);
-	GLhandleARB loadShaderFile(const std::string& filename, S32 & shader_level, GLenum type, boost::unordered_map<std::string, std::string>* defines = NULL, S32 texture_index_channels = -1);
+	GLhandleARB loadShaderFile(const std::string& filename, S32 & shader_level, GLenum type, std::unordered_map<std::string, std::string>* defines = NULL, S32 texture_index_channels = -1);
 
 	// Implemented in the application to actually point to the shader directory.
 	virtual std::string getShaderDirPrefix(void) = 0; // Pure Virtual
diff --git a/indra/llrender/lltexture.h b/indra/llrender/lltexture.h
index 41481fb8a7..e890a5a30b 100644
--- a/indra/llrender/lltexture.h
+++ b/indra/llrender/lltexture.h
@@ -42,7 +42,7 @@ class LLFontGL ;
 //
 //this is an abstract class as the parent for the class LLGLTexture
 //
-class LLTexture : public virtual LLRefCount, public LLTrace::MemTrackable<LLTexture>
+class LLTexture : public virtual LLRefCount
 {
 	friend class LLTexUnit ;
 	friend class LLFontGL ;
@@ -52,7 +52,6 @@ protected:
 
 public:
 	LLTexture()
-	:	LLTrace::MemTrackable<LLTexture>("LLTexture")
 	{}
 
 	//
@@ -67,11 +66,9 @@ public:
 	virtual S32	       getWidth(S32 discard_level = -1) const;
 	virtual S32	       getHeight(S32 discard_level = -1) const;
 	virtual bool       isActiveFetching();
+    virtual LLImageGL* getGLTexture() const;
 
 private:
-	//note: do not make this function public.
-	virtual LLImageGL* getGLTexture() const;
-
 	virtual void updateBindStatsForTester();
 };
 #endif
diff --git a/indra/llrender/llvertexbuffer.cpp b/indra/llrender/llvertexbuffer.cpp
index 93967b128d..954b3d5fa8 100644
--- a/indra/llrender/llvertexbuffer.cpp
+++ b/indra/llrender/llvertexbuffer.cpp
@@ -68,8 +68,8 @@ const U32 LL_VBO_POOL_MAX_SEED_SIZE = 256*1024;
 
 U32 vbo_block_size(U32 size)
 { //what block size will fit size?
-    U32 mod = size % LL_VBO_BLOCK_SIZE;
-    return mod == 0 ? size : size + (LL_VBO_BLOCK_SIZE-mod);
+	U32 mod = size % LL_VBO_BLOCK_SIZE;
+	return mod == 0 ? size : size + (LL_VBO_BLOCK_SIZE-mod);
 }
 
 U32 vbo_block_index(U32 size)
@@ -93,6 +93,8 @@ LLVBOPool LLVertexBuffer::sDynamicIBOPool(GL_DYNAMIC_DRAW_ARB, GL_ELEMENT_ARRAY_
 
 U32 LLVBOPool::sBytesPooled = 0;
 U32 LLVBOPool::sIndexBytesPooled = 0;
+U32 LLVBOPool::sNameIdx = 0;
+U32 LLVBOPool::sNamePool[1024];
 
 std::list<U32> LLVertexBuffer::sAvailableVAOName;
 U32 LLVertexBuffer::sCurVAOName = 1;
@@ -122,15 +124,20 @@ bool LLVertexBuffer::sPreferStreamDraw = false;
 
 U32 LLVBOPool::genBuffer()
 {
-	U32 ret = 0;
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX
 
-	glGenBuffersARB(1, &ret);
+	if (sNameIdx == 0)
+	{
+		glGenBuffersARB(1024, sNamePool);
+		sNameIdx = 1024;
+	}
 	
-	return ret;
+	return sNamePool[--sNameIdx];
 }
 
 void LLVBOPool::deleteBuffer(U32 name)
 {
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX
 	if (gGLManager.mInited)
 	{
 		LLVertexBuffer::unbind();
@@ -148,112 +155,113 @@ LLVBOPool::LLVBOPool(U32 vboUsage, U32 vboType)
 : mUsage(vboUsage), mType(vboType), mMissCountDirty(true)
 {
     mFreeList.resize(LL_VBO_POOL_SEED_COUNT);
-    mMissCount.resize(LL_VBO_POOL_SEED_COUNT);
+	mMissCount.resize(LL_VBO_POOL_SEED_COUNT);
 	std::fill(mMissCount.begin(), mMissCount.end(), 0);
 }
 
-volatile U8 *LLVBOPool::allocate(U32 &name, U32 size, bool for_seed)
+U8* LLVBOPool::allocate(U32& name, U32 size, bool for_seed)
 {
-    llassert(vbo_block_size(size) == size);
-
-    volatile U8 *ret = NULL;
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX
+	llassert(vbo_block_size(size) == size);
+	
+	U8* ret = NULL;
 
-    U32 i = vbo_block_index(size);
+	U32 i = vbo_block_index(size);
 
-    if (mFreeList.size() <= i)
-    {
-        mFreeList.resize(i + 1);
-    }
-
-    if (mFreeList[i].empty() || for_seed)
-    {
-        // make a new buffer
-        name = genBuffer();
+	if (mFreeList.size() <= i)
+	{
+		mFreeList.resize(i+1);
+	}
 
-        glBindBufferARB(mType, name);
+	if (mFreeList[i].empty() || for_seed)
+	{
+		//make a new buffer
+		name = genBuffer();
+		
+		glBindBufferARB(mType, name);
 
-        if (!for_seed && i < LL_VBO_POOL_SEED_COUNT)
-        {  // record this miss
-            mMissCount[i]++;
+		if (!for_seed && i < LL_VBO_POOL_SEED_COUNT)
+		{ //record this miss
+			mMissCount[i]++;	
             mMissCountDirty = true;  // signal to ::seedPool()
-        }
+		}
 
-        if (mType == GL_ARRAY_BUFFER_ARB)
-        {
-            LLVertexBuffer::sAllocatedBytes += size;
-        }
-        else
-        {
-            LLVertexBuffer::sAllocatedIndexBytes += size;
-        }
+		if (mType == GL_ARRAY_BUFFER_ARB)
+		{
+			LLVertexBuffer::sAllocatedBytes += size;
+		}
+		else
+		{
+			LLVertexBuffer::sAllocatedIndexBytes += size;
+		}
 
-        if (LLVertexBuffer::sDisableVBOMapping || mUsage != GL_DYNAMIC_DRAW_ARB)
-        {
-            glBufferDataARB(mType, size, 0, mUsage);
-            if (mUsage != GL_DYNAMIC_COPY_ARB)
-            {  // data will be provided by application
-                ret = (U8 *) ll_aligned_malloc<64>(size);
-                if (!ret)
-                {
+		if (LLVertexBuffer::sDisableVBOMapping || mUsage != GL_DYNAMIC_DRAW_ARB)
+		{
+			glBufferDataARB(mType, size, 0, mUsage);
+			if (mUsage != GL_DYNAMIC_COPY_ARB)
+			{ //data will be provided by application
+				ret = (U8*) ll_aligned_malloc<64>(size);
+				if (!ret)
+				{
                     LL_ERRS()
                         << "Failed to allocate " << size << " bytes for LLVBOPool buffer " << name << "." << LL_NEWLINE
                         << "Free list size: "
                         << mFreeList.size()  // this happens if we are out of memory so a solution might be to clear some from freelist
-                        << " Allocated Bytes: " << LLVertexBuffer::sAllocatedBytes
+							  << " Allocated Bytes: " << LLVertexBuffer::sAllocatedBytes
                         << " Allocated Index Bytes: " << LLVertexBuffer::sAllocatedIndexBytes << " Pooled Bytes: " << sBytesPooled
                         << " Pooled Index Bytes: " << sIndexBytesPooled << LL_ENDL;
-                }
-            }
-        }
-        else
-        {  // always use a true hint of static draw when allocating non-client-backed buffers
-            glBufferDataARB(mType, size, 0, GL_STATIC_DRAW_ARB);
-        }
+				}
+			}
+		}
+		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);
+
+		if (for_seed)
+		{ //put into pool for future use
+			llassert(mFreeList.size() > i);
 
-        glBindBufferARB(mType, 0);
-
-        if (for_seed)
-        {  // put into pool for future use
-            llassert(mFreeList.size() > i);
-
-            Record rec;
-            rec.mGLName     = name;
-            rec.mClientData = ret;
-
-            if (mType == GL_ARRAY_BUFFER_ARB)
-            {
-                sBytesPooled += size;
-            }
-            else
-            {
-                sIndexBytesPooled += size;
-            }
-            mFreeList[i].push_back(rec);
+			Record rec;
+			rec.mGLName = name;
+			rec.mClientData = ret;
+	
+			if (mType == GL_ARRAY_BUFFER_ARB)
+			{
+				sBytesPooled += size;
+			}
+			else
+			{
+				sIndexBytesPooled += size;
+			}
+			mFreeList[i].push_back(rec);
             mMissCountDirty = true;  // signal to ::seedPool()
-        }
-    }
-    else
-    {
-        name = mFreeList[i].front().mGLName;
-        ret  = mFreeList[i].front().mClientData;
+		}
+	}
+	else
+	{
+		name = mFreeList[i].front().mGLName;
+		ret = mFreeList[i].front().mClientData;
 
-        if (mType == GL_ARRAY_BUFFER_ARB)
-        {
-            sBytesPooled -= size;
-        }
-        else
-        {
-            sIndexBytesPooled -= size;
-        }
+		if (mType == GL_ARRAY_BUFFER_ARB)
+		{
+			sBytesPooled -= size;
+		}
+		else
+		{
+			sIndexBytesPooled -= size;
+		}
 
-        mFreeList[i].pop_front();
+		mFreeList[i].pop_front();
         mMissCountDirty = true;  // signal to ::seedPool()
-    }
+	}
 
-    return ret;
+	return ret;
 }
 
-void LLVBOPool::release(U32 name, volatile U8* buffer, U32 size)
+void LLVBOPool::release(U32 name, U8* buffer, U32 size)
 {
 	llassert(vbo_block_size(size) == size);
 
@@ -272,25 +280,26 @@ void LLVBOPool::release(U32 name, volatile U8* buffer, U32 size)
 
 void LLVBOPool::seedPool()
 {
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX
     if (mMissCountDirty)
     {
-        U32 dummy_name = 0;
+	U32 dummy_name = 0;
         U32 size       = LL_VBO_BLOCK_SIZE;
 
-        for (U32 i = 0; i < LL_VBO_POOL_SEED_COUNT; i++)
-        {
-            if (mMissCount[i] > mFreeList[i].size())
-            {
-                S32 count = mMissCount[i] - mFreeList[i].size();
-                for (U32 j = 0; j < count; ++j)
-                {
-                    allocate(dummy_name, size, true);
-                }
-            }
+	for (U32 i = 0; i < LL_VBO_POOL_SEED_COUNT; i++)
+	{
+		if (mMissCount[i] > mFreeList[i].size())
+		{ 
+			S32 count = mMissCount[i] - mFreeList[i].size();
+			for (U32 j = 0; j < count; ++j)
+			{
+				allocate(dummy_name, size, true);
+			}
+		}
             size += LL_VBO_BLOCK_SIZE;
         }
         mMissCountDirty = false;
-    }
+	}
 }
 
 void LLVBOPool::cleanup()
@@ -413,6 +422,7 @@ void LLVertexBuffer::releaseVAOName(U32 name)
 //static
 void LLVertexBuffer::seedPools()
 {
+	LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX
 	sStreamVBOPool.seedPool();
 	sDynamicVBOPool.seedPool();
 	sDynamicCopyVBOPool.seedPool();
@@ -432,8 +442,6 @@ void LLVertexBuffer::setupClientArrays(U32 data_mask)
 			data_mask = data_mask & ~MAP_TEXTURE_INDEX;
 		}
 
-		if (LLGLSLShader::sNoFixedFunction)
-		{
 			for (U32 i = 0; i < TYPE_MAX; ++i)
 			{
 				S32 loc = i;
@@ -455,176 +463,29 @@ void LLVertexBuffer::setupClientArrays(U32 data_mask)
 					}
 				}
 			}
-		}
-		else
-		{
-
-			static const GLenum array[] =
-			{
-				GL_VERTEX_ARRAY,
-				GL_NORMAL_ARRAY,
-				GL_TEXTURE_COORD_ARRAY,
-				GL_COLOR_ARRAY,
-			};
-
-			static const 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)
-							{
-								gFailLog << "Bad client state! " << array[i] << " disabled." << std::endl;
-							}
-							else
-							{
-								LL_ERRS() << "Bad client state! " << array[i] << " disabled." << LL_ENDL;
-							}
-						}
-					}
-				}
-				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)
-						{
-							gFailLog << "Bad client state! " << array[i] << " enabled." << std::endl;
-						}
-						else
-						{
-							LL_ERRS() << "Bad client state! " << array[i] << " enabled." << LL_ENDL;
-						}
-					}
-				}
-			}
-		
-			static const 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 (sLastMask & MAP_TANGENT)
-			{
-				if (!(data_mask & MAP_TANGENT))
-				{
-					glClientActiveTextureARB(GL_TEXTURE2_ARB);
-					glDisableClientState(GL_TEXTURE_COORD_ARRAY);
-					glClientActiveTextureARB(GL_TEXTURE0_ARB);
-				}
-			}
-			else if (data_mask & MAP_TANGENT)
-			{
-				glClientActiveTextureARB(GL_TEXTURE2_ARB);
-				glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-				glClientActiveTextureARB(GL_TEXTURE0_ARB);
-			}
-		}
 				
 		sLastMask = data_mask;
 	}
 }
 
 //static
-static LLTrace::BlockTimerStatHandle FTM_VB_DRAW_ARRAYS("drawArrays");
-void LLVertexBuffer::drawArrays(U32 mode, const std::vector<LLVector3>& pos, const std::vector<LLVector3>& norm)
+void LLVertexBuffer::drawArrays(U32 mode, const std::vector<LLVector3>& pos)
 {
-	LL_RECORD_BLOCK_TIME(FTM_VB_DRAW_ARRAYS);
-	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
-	gGL.syncMatrices();
-
-	U32 count = pos.size();
-	
-	llassert(norm.size() >= pos.size());
-	llassert(count > 0);
-
-	if( count == 0 )
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
+    gGL.begin(mode);
+    for (auto& v : pos)
 	{
-		LL_WARNS() << "Called drawArrays with 0 vertices" << LL_ENDL;
-		return;
-	}
-
-	if( norm.size() < pos.size() )
-	{
-		LL_WARNS() << "Called drawArrays with #" << norm.size() << " normals and #" << pos.size() << " vertices" << LL_ENDL;
-		return;
+        gGL.vertex3fv(v.mV);
 	}
-
-	unbind();
-	
-	setupClientArrays(MAP_VERTEX | MAP_NORMAL);
-
-	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);
+    gGL.end();
+    gGL.flush();
 		}
-		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);
-	}
-	LLGLSLShader::startProfile();
-	glDrawArrays(sGLMode[mode], 0, count);
-	LLGLSLShader::stopProfile(count, mode);
-}
 
 //static
 void LLVertexBuffer::drawElements(U32 mode, const LLVector4a* pos, const LLVector2* tc, S32 num_indices, const U16* indicesp)
 {
-	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
+	llassert(LLGLSLShader::sCurBoundShaderPtr != NULL);
 
 	gGL.syncMatrices();
 
@@ -636,28 +497,27 @@ void LLVertexBuffer::drawElements(U32 mode, const LLVector4a* pos, const LLVecto
 
 	unbind();
 	
-	setupClientArrays(mask);
+    gGL.begin(mode);
 
-	if (LLGLSLShader::sNoFixedFunction)
+    if (tc != nullptr)
 	{
-		S32 loc = LLVertexBuffer::TYPE_VERTEX;
-		glVertexAttribPointerARB(loc, 3, GL_FLOAT, GL_FALSE, 16, pos);
-
-		if (tc)
+        for (int i = 0; i < num_indices; ++i)
 		{
-			loc = LLVertexBuffer::TYPE_TEXCOORD0;
-			glVertexAttribPointerARB(loc, 2, GL_FLOAT, GL_FALSE, 0, tc);
+            U16 idx = indicesp[i];
+            gGL.texCoord2fv(tc[idx].mV);
+            gGL.vertex3fv(pos[idx].getF32ptr());
 		}
 	}
 	else
 	{
-		glTexCoordPointer(2, GL_FLOAT, 0, tc);
-		glVertexPointer(3, GL_FLOAT, 16, pos);
+        for (int i = 0; i < num_indices; ++i)
+        {
+            U16 idx = indicesp[i];
+            gGL.vertex3fv(pos[idx].getF32ptr());
 	}
-
-	LLGLSLShader::startProfile();
-	glDrawElements(sGLMode[mode], num_indices, GL_UNSIGNED_SHORT, indicesp);
-	LLGLSLShader::stopProfile(num_indices, mode);
+}
+    gGL.end();
+    gGL.flush();
 }
 
 void LLVertexBuffer::validateRange(U32 start, U32 end, U32 count, U32 indices_offset) const
@@ -722,7 +582,7 @@ void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indi
 	gGL.syncMatrices();
 
 	llassert(mNumVerts >= 0);
-	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+	llassert(LLGLSLShader::sCurBoundShaderPtr != NULL);
 
 	if (mGLArray)
 	{
@@ -765,6 +625,7 @@ void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indi
 
 	stop_glerror();
 	LLGLSLShader::startProfile();
+    LL_PROFILER_GPU_ZONEC( "gl.DrawRangeElements", 0xFFFF00 )
 	glDrawRangeElements(sGLMode[mode], start, end, count, GL_UNSIGNED_SHORT, 
 		idx);
 	LLGLSLShader::stopProfile(count, mode);
@@ -775,9 +636,21 @@ void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indi
 	placeFence();
 }
 
+void LLVertexBuffer::drawRangeFast(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const
+{
+    mMappable = false;
+    gGL.syncMatrices();
+
+    U16* idx = ((U16*)getIndicesPointer()) + indices_offset;
+
+    LL_PROFILER_GPU_ZONEC("gl.DrawRangeElements", 0xFFFF00)
+        glDrawRangeElements(sGLMode[mode], start, end, count, GL_UNSIGNED_SHORT,
+            idx);
+}
+
 void LLVertexBuffer::draw(U32 mode, U32 count, U32 indices_offset) const
 {
-	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+	llassert(LLGLSLShader::sCurBoundShaderPtr != NULL);
 	mMappable = false;
 	gGL.syncMatrices();
 
@@ -816,6 +689,7 @@ void LLVertexBuffer::draw(U32 mode, U32 count, U32 indices_offset) const
 
 	stop_glerror();
 	LLGLSLShader::startProfile();
+    LL_PROFILER_GPU_ZONEC( "gl.DrawElements", 0xA0FFA0 )
 	glDrawElements(sGLMode[mode], count, GL_UNSIGNED_SHORT,
 		((U16*) getIndicesPointer()) + indices_offset);
 	LLGLSLShader::stopProfile(count, mode);
@@ -823,13 +697,15 @@ void LLVertexBuffer::draw(U32 mode, U32 count, U32 indices_offset) const
 	placeFence();
 }
 
-static LLTrace::BlockTimerStatHandle FTM_GL_DRAW_ARRAYS("GL draw arrays");
+
 void LLVertexBuffer::drawArrays(U32 mode, U32 first, U32 count) const
 {
-	llassert(!LLGLSLShader::sNoFixedFunction || LLGLSLShader::sCurBoundShaderPtr != NULL);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
+    llassert(LLGLSLShader::sCurBoundShaderPtr != NULL);
 	mMappable = false;
 	gGL.syncMatrices();
 	
+#ifndef LL_RELEASE_FOR_DOWNLOAD
 	llassert(mNumVerts >= 0);
 	if (first >= (U32) mNumVerts ||
 	    first + count > (U32) mNumVerts)
@@ -857,16 +733,14 @@ void LLVertexBuffer::drawArrays(U32 mode, U32 first, U32 count) const
 		LL_ERRS() << "Invalid draw mode: " << mode << LL_ENDL;
 		return;
 	}
+#endif
 
+    LLGLSLShader::startProfile();
 	{
-		LL_RECORD_BLOCK_TIME(FTM_GL_DRAW_ARRAYS);
-		stop_glerror();
-		LLGLSLShader::startProfile();
-		stop_glerror();
+        LL_PROFILER_GPU_ZONEC("gl.DrawArrays", 0xFF4040)
 		glDrawArrays(sGLMode[mode], first, count);
-		stop_glerror();
+    }
 		LLGLSLShader::stopProfile(count, mode);
-	}
 
 	stop_glerror();
 	placeFence();
@@ -971,8 +845,7 @@ S32 LLVertexBuffer::determineUsage(S32 usage)
 }
 
 LLVertexBuffer::LLVertexBuffer(U32 typemask, S32 usage) 
-:	LLTrace::MemTrackable<LLVertexBuffer>("LLVertexBuffer"),
-	LLRefCount(),
+:	LLRefCount(),
 
 	mNumVerts(0),
 	mNumIndices(0),
@@ -1117,9 +990,7 @@ void LLVertexBuffer::waitFence() const
 
 void LLVertexBuffer::genBuffer(U32 size)
 {
-	disclaimMem(mSize);
 	mSize = vbo_block_size(size);
-	claimMem(mSize);
 
 	if (mUsage == GL_STREAM_DRAW_ARB)
 	{
@@ -1215,9 +1086,7 @@ bool LLVertexBuffer::createGLBuffer(U32 size)
 		static int gl_buffer_idx = 0;
 		mGLBuffer = ++gl_buffer_idx;
 		mMappedData = (U8*)ll_aligned_malloc_16(size);
-		disclaimMem(mSize);
 		mSize = size;
-		claimMem(mSize);
 	}
 
 	if (!mMappedData)
@@ -1387,8 +1256,6 @@ bool LLVertexBuffer::allocateBuffer(S32 nverts, S32 nindices, bool create)
 	return success;
 }
 
-static LLTrace::BlockTimerStatHandle FTM_SETUP_VERTEX_ARRAY("Setup VAO");
-
 void LLVertexBuffer::setupVertexArray()
 {
 	if (!mGLArray)
@@ -1396,7 +1263,7 @@ void LLVertexBuffer::setupVertexArray()
 		return;
 	}
 
-	LL_RECORD_BLOCK_TIME(FTM_SETUP_VERTEX_ARRAY);
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 #if GL_ARB_vertex_array_object
 	glBindVertexArray(mGLArray);
 #endif
@@ -1569,12 +1436,11 @@ bool expand_region(LLVertexBuffer::MappedRegion& region, S32 index, S32 count)
 	return true;
 }
 
-static LLTrace::BlockTimerStatHandle FTM_VBO_MAP_BUFFER_RANGE("VBO Map Range");
-static LLTrace::BlockTimerStatHandle FTM_VBO_MAP_BUFFER("VBO Map");
 
 // Map for data access
-volatile U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range)
+U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 	bindGLBuffer(true);
 	if (mFinal)
 	{
@@ -1634,14 +1500,13 @@ volatile U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, boo
 			}
 			else
 			{
-				volatile U8* src = NULL;
+				U8* src = NULL;
 				waitFence();
 				if (gGLManager.mHasMapBufferRange)
 				{
 					if (map_range)
 					{
 #ifdef GL_ARB_map_buffer_range
-						LL_RECORD_BLOCK_TIME(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, 
@@ -1665,7 +1530,6 @@ volatile U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, boo
 							}
 						}
 
-						LL_RECORD_BLOCK_TIME(FTM_VBO_MAP_BUFFER);
 						src = (U8*) glMapBufferRange(GL_ARRAY_BUFFER_ARB, 0, mSize, 
 							GL_MAP_WRITE_BIT | 
 							GL_MAP_FLUSH_EXPLICIT_BIT);
@@ -1695,7 +1559,7 @@ volatile U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, boo
 
 				llassert(src != NULL);
 
-				mMappedData = LL_NEXT_ALIGNED_ADDRESS<volatile U8>(src);
+				mMappedData = LL_NEXT_ALIGNED_ADDRESS<U8>(src);
 				mAlignedOffset = mMappedData - src;
 			
 				stop_glerror();
@@ -1751,11 +1615,9 @@ volatile U8* LLVertexBuffer::mapVertexBuffer(S32 type, S32 index, S32 count, boo
 }
 
 
-static LLTrace::BlockTimerStatHandle FTM_VBO_MAP_INDEX_RANGE("IBO Map Range");
-static LLTrace::BlockTimerStatHandle FTM_VBO_MAP_INDEX("IBO Map");
-
-volatile U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
+U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 	bindGLIndices(true);
 	if (mFinal)
 	{
@@ -1823,14 +1685,13 @@ volatile U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range
 			}
 			else
 			{
-				volatile U8* src = NULL;
+				U8* src = NULL;
 				waitFence();
 				if (gGLManager.mHasMapBufferRange)
 				{
 					if (map_range)
 					{
 #ifdef GL_ARB_map_buffer_range
-						LL_RECORD_BLOCK_TIME(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, 
@@ -1842,7 +1703,6 @@ volatile U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range
 					else
 					{
 #ifdef GL_ARB_map_buffer_range
-						LL_RECORD_BLOCK_TIME(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);
@@ -1866,7 +1726,6 @@ volatile U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range
 				}
 				else
 				{
-					LL_RECORD_BLOCK_TIME(FTM_VBO_MAP_INDEX);
 					map_range = false;
 					src = (U8*) glMapBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
 				}
@@ -1917,13 +1776,6 @@ volatile U8* LLVertexBuffer::mapIndexBuffer(S32 index, S32 count, bool map_range
 	}
 }
 
-static LLTrace::BlockTimerStatHandle FTM_VBO_UNMAP("VBO Unmap");
-static LLTrace::BlockTimerStatHandle FTM_VBO_FLUSH_RANGE("Flush VBO Range");
-
-
-static LLTrace::BlockTimerStatHandle FTM_IBO_UNMAP("IBO Unmap");
-static LLTrace::BlockTimerStatHandle FTM_IBO_FLUSH_RANGE("Flush IBO Range");
-
 void LLVertexBuffer::unmapBuffer()
 {
 	if (!useVBOs())
@@ -1932,10 +1784,10 @@ void LLVertexBuffer::unmapBuffer()
 	}
 
 	bool updated_all = false;
-
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 	if (mMappedData && mVertexLocked)
 	{
-		LL_RECORD_BLOCK_TIME(FTM_VBO_UNMAP);
+        LL_PROFILE_ZONE_NAMED_CATEGORY_VERTEX("unmapBuffer - vertex");
 		bindGLBuffer(true);
 		updated_all = mIndexLocked; //both vertex and index buffers done updating
 
@@ -1982,7 +1834,7 @@ void LLVertexBuffer::unmapBuffer()
 			{
 				if (!mMappedVertexRegions.empty())
 				{
-					stop_glerror();
+                    LL_PROFILE_ZONE_NAMED_CATEGORY_VERTEX("unmapBuffer - flush vertex");
 					for (U32 i = 0; i < mMappedVertexRegions.size(); ++i)
 					{
 						const MappedRegion& region = mMappedVertexRegions[i];
@@ -1990,7 +1842,6 @@ void LLVertexBuffer::unmapBuffer()
 						S32 length = sTypeSize[region.mType]*region.mCount;
 						if (gGLManager.mHasMapBufferRange)
 						{
-							LL_RECORD_BLOCK_TIME(FTM_VBO_FLUSH_RANGE);
 #ifdef GL_ARB_map_buffer_range
 							glFlushMappedBufferRange(GL_ARRAY_BUFFER_ARB, offset, length);
 #endif
@@ -2001,7 +1852,6 @@ void LLVertexBuffer::unmapBuffer()
 							glFlushMappedBufferRangeAPPLE(GL_ARRAY_BUFFER_ARB, offset, length);
 #endif
 						}
-						stop_glerror();
 					}
 
 					mMappedVertexRegions.clear();
@@ -2020,7 +1870,7 @@ void LLVertexBuffer::unmapBuffer()
 	
 	if (mMappedIndexData && mIndexLocked)
 	{
-		LL_RECORD_BLOCK_TIME(FTM_IBO_UNMAP);
+        LL_PROFILE_ZONE_NAMED_CATEGORY_VERTEX("unmapBuffer - index");
 		bindGLIndices();
 		if(!mMappable)
 		{
@@ -2066,12 +1916,12 @@ void LLVertexBuffer::unmapBuffer()
 				{
 					for (U32 i = 0; i < mMappedIndexRegions.size(); ++i)
 					{
+                        LL_PROFILE_ZONE_NAMED_CATEGORY_VERTEX("unmapBuffer - flush index");
 						const MappedRegion& region = mMappedIndexRegions[i];
 						S32 offset = region.mIndex >= 0 ? sizeof(U16)*region.mIndex : 0;
 						S32 length = sizeof(U16)*region.mCount;
 						if (gGLManager.mHasMapBufferRange)
 						{
-							LL_RECORD_BLOCK_TIME(FTM_IBO_FLUSH_RANGE);
 #ifdef GL_ARB_map_buffer_range
 							glFlushMappedBufferRange(GL_ELEMENT_ARRAY_BUFFER_ARB, offset, length);
 #endif
@@ -2090,9 +1940,8 @@ void LLVertexBuffer::unmapBuffer()
 					mMappedIndexRegions.clear();
 				}
 			}
-			stop_glerror();
+			
 			glUnmapBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB);
-			stop_glerror();
 
 			mMappedIndexData = NULL;
 		}
@@ -2118,7 +1967,7 @@ template <class T,S32 type> struct VertexBufferStrider
 	{
 		if (type == LLVertexBuffer::TYPE_INDEX)
 		{
-			volatile U8* ptr = vbo.mapIndexBuffer(index, count, map_range);
+			U8* ptr = vbo.mapIndexBuffer(index, count, map_range);
 
 			if (ptr == NULL)
 			{
@@ -2134,7 +1983,7 @@ template <class T,S32 type> struct VertexBufferStrider
 		{
 			S32 stride = LLVertexBuffer::sTypeSize[type];
 
-			volatile U8* ptr = vbo.mapVertexBuffer(type, index, count, map_range);
+			U8* ptr = vbo.mapVertexBuffer(type, index, count, map_range);
 
 			if (ptr == NULL)
 			{
@@ -2215,13 +2064,12 @@ bool LLVertexBuffer::getClothWeightStrider(LLStrider<LLVector4>& strider, S32 in
 
 //----------------------------------------------------------------------------
 
-static LLTrace::BlockTimerStatHandle FTM_BIND_GL_ARRAY("Bind Array");
 bool LLVertexBuffer::bindGLArray()
 {
 	if (mGLArray && sGLRenderArray != mGLArray)
 	{
 		{
-			LL_RECORD_BLOCK_TIME(FTM_BIND_GL_ARRAY);
+            LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 #if GL_ARB_vertex_array_object
 			glBindVertexArray(mGLArray);
 #endif
@@ -2238,8 +2086,6 @@ bool LLVertexBuffer::bindGLArray()
 	return false;
 }
 
-static LLTrace::BlockTimerStatHandle FTM_BIND_GL_BUFFER("Bind Buffer");
-
 bool LLVertexBuffer::bindGLBuffer(bool force_bind)
 {
 	bindGLArray();
@@ -2248,8 +2094,7 @@ bool LLVertexBuffer::bindGLBuffer(bool force_bind)
 
 	if (useVBOs() && (force_bind || (mGLBuffer && (mGLBuffer != sGLRenderBuffer || !sVBOActive))))
 	{
-		//LL_RECORD_BLOCK_TIME(FTM_BIND_GL_BUFFER);
-		
+        LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 		glBindBufferARB(GL_ARRAY_BUFFER_ARB, mGLBuffer);
 		sGLRenderBuffer = mGLBuffer;
 		sBindCount++;
@@ -2263,16 +2108,29 @@ bool LLVertexBuffer::bindGLBuffer(bool force_bind)
 	return ret;
 }
 
-static LLTrace::BlockTimerStatHandle FTM_BIND_GL_INDICES("Bind Indices");
+bool LLVertexBuffer::bindGLBufferFast()
+{
+    if (mGLBuffer != sGLRenderBuffer || !sVBOActive)
+    {
+        glBindBufferARB(GL_ARRAY_BUFFER_ARB, mGLBuffer);
+        sGLRenderBuffer = mGLBuffer;
+        sBindCount++;
+        sVBOActive = true;
+
+        return true;
+    }
+
+    return false;
+}
 
 bool LLVertexBuffer::bindGLIndices(bool force_bind)
 {
+    LL_PROFILE_ZONE_SCOPED_CATEGORY_VERTEX;
 	bindGLArray();
 
 	bool ret = false;
 	if (useVBOs() && (force_bind || (mGLIndices && (mGLIndices != sGLRenderIndices || !sIBOActive))))
 	{
-		LL_RECORD_BLOCK_TIME(FTM_BIND_GL_INDICES);
 		/*if (sMapped)
 		{
 			LL_ERRS() << "VBO bound while another VBO mapped!" << LL_ENDL;
@@ -2288,6 +2146,21 @@ bool LLVertexBuffer::bindGLIndices(bool force_bind)
 	return ret;
 }
 
+bool LLVertexBuffer::bindGLIndicesFast()
+{
+    if (mGLIndices != sGLRenderIndices || !sIBOActive)
+    {
+        glBindBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, mGLIndices);
+        sGLRenderIndices = mGLIndices;
+        sBindCount++;
+        sIBOActive = true;
+        
+        return true;
+    }
+
+    return false;
+}
+
 void LLVertexBuffer::flush()
 {
 	if (useVBOs())
@@ -2478,11 +2351,39 @@ void LLVertexBuffer::setBuffer(U32 data_mask)
 	}
 }
 
+void LLVertexBuffer::setBufferFast(U32 data_mask)
+{
+    if (useVBOs())
+    {
+        //set up pointers if the data mask is different ...
+        bool setup = (sLastMask != data_mask);
+
+        const bool bindBuffer = bindGLBufferFast();
+        const bool bindIndices = bindGLIndicesFast();
+
+        setup = setup || bindBuffer || bindIndices;
+        
+        setupClientArrays(data_mask);
+
+        if (data_mask && setup)
+        {
+            setupVertexBufferFast(data_mask);
+            sSetCount++;
+        }
+    }
+    else
+    {
+        //fallback to slow path when not using VBOs
+        setBuffer(data_mask);
+    }
+}
+
+
 // virtual (default)
 void LLVertexBuffer::setupVertexBuffer(U32 data_mask)
 {
 	stop_glerror();
-	volatile U8* base = useVBOs() ? (U8*) mAlignedOffset : mMappedData;
+	U8* base = useVBOs() ? (U8*) mAlignedOffset : mMappedData;
 
 	if (gDebugGL && ((data_mask & mTypeMask) != data_mask))
 	{
@@ -2497,8 +2398,6 @@ void LLVertexBuffer::setupVertexBuffer(U32 data_mask)
 		LL_ERRS() << "LLVertexBuffer::setupVertexBuffer missing required components for supplied data mask." << LL_ENDL;
 	}
 
-	if (LLGLSLShader::sNoFixedFunction)
-	{
 		if (data_mask & MAP_NORMAL)
 		{
 			S32 loc = TYPE_NORMAL;
@@ -2587,54 +2486,103 @@ void LLVertexBuffer::setupVertexBuffer(U32 data_mask)
 			void* ptr = (void*)(base + mOffsets[TYPE_VERTEX]);
 			glVertexAttribPointerARB(loc, 3,GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_VERTEX], ptr);
 		}	
+
+	llglassertok();
 	}	
-	else
+
+void LLVertexBuffer::setupVertexBufferFast(U32 data_mask)
 	{
+    U8* base = (U8*)mAlignedOffset;
+
 		if (data_mask & MAP_NORMAL)
 		{
-			glNormalPointer(GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_NORMAL], (void*)(base + mOffsets[TYPE_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)
 		{
-			glClientActiveTextureARB(GL_TEXTURE3_ARB);
-			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD3], (void*)(base + mOffsets[TYPE_TEXCOORD3]));
-			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+        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)
 		{
-			glClientActiveTextureARB(GL_TEXTURE2_ARB);
-			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD2], (void*)(base + mOffsets[TYPE_TEXCOORD2]));
-			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+        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)
 		{
-			glClientActiveTextureARB(GL_TEXTURE1_ARB);
-			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD1], (void*)(base + mOffsets[TYPE_TEXCOORD1]));
-			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+        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_TANGENT)
 		{
-			glClientActiveTextureARB(GL_TEXTURE2_ARB);
-			glTexCoordPointer(4,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TANGENT], (void*)(base + mOffsets[TYPE_TANGENT]));
-			glClientActiveTextureARB(GL_TEXTURE0_ARB);
+        S32 loc = TYPE_TANGENT;
+        void* ptr = (void*)(base + mOffsets[TYPE_TANGENT]);
+        glVertexAttribPointerARB(loc, 4, GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_TANGENT], ptr);
 		}
 		if (data_mask & MAP_TEXCOORD0)
 		{
-			glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD0], (void*)(base + mOffsets[TYPE_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)
 		{
-			glColorPointer(4, GL_UNSIGNED_BYTE, LLVertexBuffer::sTypeSize[TYPE_COLOR], (void*)(base + mOffsets[TYPE_COLOR]));
+        S32 loc = TYPE_COLOR;
+        //bind emissive instead of color pointer if emissive is present
+        void* ptr = (data_mask & MAP_EMISSIVE) ? (void*)(base + mOffsets[TYPE_EMISSIVE]) : (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_COLOR))
+        { //map emissive to color channel when color is not also being bound to avoid unnecessary shader swaps
+            loc = TYPE_COLOR;
+            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)
+    {
+#if !LL_DARWIN
+        S32 loc = TYPE_TEXTURE_INDEX;
+        void* ptr = (void*)(base + mOffsets[TYPE_VERTEX] + 12);
+        glVertexAttribIPointer(loc, 1, GL_UNSIGNED_INT, LLVertexBuffer::sTypeSize[TYPE_VERTEX], ptr);
+#endif
+    }
 		if (data_mask & MAP_VERTEX)
 		{
-			glVertexPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_VERTEX], (void*)(base + 0));
+        S32 loc = TYPE_VERTEX;
+        void* ptr = (void*)(base + mOffsets[TYPE_VERTEX]);
+        glVertexAttribPointerARB(loc, 3, GL_FLOAT, GL_FALSE, LLVertexBuffer::sTypeSize[TYPE_VERTEX], ptr);
 		}	
 	}
 
-	llglassertok();
-}
-
 LLVertexBuffer::MappedRegion::MappedRegion(S32 type, S32 index, S32 count)
 : mType(type), mIndex(index), mCount(count)
 { 
diff --git a/indra/llrender/llvertexbuffer.h b/indra/llrender/llvertexbuffer.h
index 649fbb782c..3c52b30570 100644
--- a/indra/llrender/llvertexbuffer.h
+++ b/indra/llrender/llvertexbuffer.h
@@ -64,10 +64,10 @@ public:
 	const U32 mType;
 
 	//size MUST be a power of 2
-	volatile U8* allocate(U32& name, U32 size, bool for_seed = false);
+	U8* allocate(U32& name, U32 size, bool for_seed = false);
 	
 	//size MUST be the size provided to allocate that returned the given name
-	void release(U32 name, volatile U8* buffer, U32 size);
+	void release(U32 name, U8* buffer, U32 size);
 	
 	//batch allocate buffers to be provided to the application on demand
 	void seedPool();
@@ -82,7 +82,7 @@ public:
 	{
 	public:
 		U32 mGLName;
-		volatile U8* mClientData;
+		U8* mClientData;
 	};
 
 	typedef std::list<Record> record_list_t;
@@ -90,13 +90,16 @@ public:
 	std::vector<U32> mMissCount;
     bool mMissCountDirty;   // flag any changes to mFreeList or mMissCount
 
+	//used to avoid calling glGenBuffers for every VBO creation
+	static U32 sNamePool[1024];
+	static U32 sNameIdx;
 };
 
 
 //============================================================================
 // base class 
 class LLPrivateMemoryPool;
-class LLVertexBuffer : public LLRefCount, public LLTrace::MemTrackable<LLVertexBuffer>
+class LLVertexBuffer : public LLRefCount
 {
 public:
 	class MappedRegion
@@ -111,8 +114,7 @@ public:
 	};
 
 	LLVertexBuffer(const LLVertexBuffer& rhs)
-	:	LLTrace::MemTrackable<LLVertexBuffer>("LLVertexBuffer"),
-		mUsage(rhs.mUsage)
+	:	mUsage(rhs.mUsage)
 	{
 		*this = rhs;
 	}
@@ -128,7 +130,7 @@ public:
 	static LLVBOPool sDynamicCopyVBOPool;
 	static LLVBOPool sStreamIBOPool;
 	static LLVBOPool sDynamicIBOPool;
-	
+
 	static std::list<U32> sAvailableVAOName;
 	static U32 sCurVAOName;
 
@@ -144,8 +146,7 @@ public:
 	static void initClass(bool use_vbo, bool no_vbo_mapping);
 	static void cleanupClass();
 	static void setupClientArrays(U32 data_mask);
-	static void pushPositions(U32 mode, const LLVector4a* pos, U32 count);
-	static void drawArrays(U32 mode, const std::vector<LLVector3>& pos, const std::vector<LLVector3>& norm);
+	static void drawArrays(U32 mode, const std::vector<LLVector3>& pos);
 	static void drawElements(U32 mode, const LLVector4a* pos, const LLVector2* tc, S32 num_indices, const U16* indicesp);
 
  	static void unbind(); //unbind any bound vertex buffer
@@ -208,13 +209,17 @@ protected:
 
 	virtual ~LLVertexBuffer(); // use unref()
 
-	virtual void setupVertexBuffer(U32 data_mask); // pure virtual, called from mapBuffer()
+	virtual void setupVertexBuffer(U32 data_mask);
+    void setupVertexBufferFast(U32 data_mask);
+
 	void setupVertexArray();
 	
 	void	genBuffer(U32 size);
 	void	genIndices(U32 size);
 	bool	bindGLBuffer(bool force_bind = false);
+    bool	bindGLBufferFast();
 	bool	bindGLIndices(bool force_bind = false);
+    bool    bindGLIndicesFast();
 	bool	bindGLArray();
 	void	releaseBuffer();
 	void	releaseIndices();
@@ -230,13 +235,15 @@ public:
 	LLVertexBuffer(U32 typemask, S32 usage);
 	
 	// map for data access
-	volatile U8*		mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range);
-	volatile U8*		mapIndexBuffer(S32 index, S32 count, bool map_range);
+	U8*		mapVertexBuffer(S32 type, S32 index, S32 count, bool map_range);
+	U8*		mapIndexBuffer(S32 index, S32 count, bool map_range);
 
 	void bindForFeedback(U32 channel, U32 type, U32 index, U32 count);
 
 	// set for rendering
 	virtual void	setBuffer(U32 data_mask); 	// calls  setupVertexBuffer() if data_mask is not 0
+    void	setBufferFast(U32 data_mask); 	// calls setupVertexBufferFast(), assumes data_mask is not 0 among other assumptions
+
 	void flush(); //flush pending data to GL memory
 	// allocate buffer
 	bool	allocateBuffer(S32 nverts, S32 nindices, bool create);
@@ -272,14 +279,14 @@ public:
 	S32 getNumVerts() const					{ return mNumVerts; }
 	S32 getNumIndices() const				{ return mNumIndices; }
 	
-	volatile U8* getIndicesPointer() const			{ return useVBOs() ? (U8*) mAlignedIndexOffset : mMappedIndexData; }
-	volatile U8* getVerticesPointer() const			{ return useVBOs() ? (U8*) mAlignedOffset : mMappedData; }
+	U8* getIndicesPointer() const			{ return useVBOs() ? (U8*) mAlignedIndexOffset : mMappedIndexData; }
+	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 mIndicesSize; }
-	volatile U8* getMappedData() const				{ return mMappedData; }
-	volatile U8* getMappedIndices() const			{ return mMappedIndexData; }
+	U8* getMappedData() const				{ return mMappedData; }
+	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; }
@@ -288,6 +295,9 @@ public:
 	void drawArrays(U32 mode, U32 offset, U32 count) const;
 	void drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const;
 
+    //implementation for inner loops that does no safety checking
+    void drawRangeFast(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const;
+
 	//for debugging, validate data in given range is valid
 	void validateRange(U32 start, U32 end, U32 count, U32 offset) const;
 
@@ -309,8 +319,8 @@ protected:
 	U32		mGLIndices;		// GL IBO handle
 	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)
+	U8* mMappedData;	// pointer to currently mapped data (NULL if unmapped)
+	U8* mMappedIndexData;	// pointer to currently mapped indices (NULL if unmapped)
 
 	U32		mMappedDataUsingVBOs : 1;
 	U32		mMappedIndexDataUsingVBOs : 1;