diff --git a/examples/06-bump/bump.cpp b/examples/06-bump/bump.cpp
index 857006f7..9128f27c 100644
--- a/examples/06-bump/bump.cpp
+++ b/examples/06-bump/bump.cpp
@@ -408,14 +408,16 @@ int _main_(int /*_argc*/, char** /*_argv*/)
 		bgfx::setViewTransform(0, view, proj);
 
 		const uint16_t instanceStride = 64;
-		const bgfx::InstanceDataBuffer* idb = bgfx::allocInstanceDataBuffer(9, instanceStride);
-		if (NULL != idb)
-		{
-			uint8_t* data = idb->data;
+		const uint16_t numInstances = 3;
 
-			// Write instance data for 3x3 cubes.
-			for (uint32_t yy = 0; yy < 3; ++yy)
+		// Write instance data for 3x3 cubes.
+		for (uint32_t yy = 0; yy < 3; ++yy)
+		{
+			const bgfx::InstanceDataBuffer* idb = bgfx::allocInstanceDataBuffer(numInstances, instanceStride);
+			if (NULL != idb)
 			{
+				uint8_t* data = idb->data;
+
 				for (uint32_t xx = 0; xx < 3; ++xx)
 				{
 					float* mtx = (float*)data;
@@ -424,43 +426,35 @@ int _main_(int /*_argc*/, char** /*_argv*/)
 					mtx[13] = -3.0f + float(yy)*3.0f;
 					mtx[14] = 0.0f;
 
-					float* color = (float*)&data[64];
-					color[0] = sin(time+float(xx)/11.0f)*0.5f+0.5f;
-					color[1] = cos(time+float(yy)/11.0f)*0.5f+0.5f;
-					color[2] = sin(time*3.0f)*0.5f+0.5f;
-					color[3] = 1.0f;
-
 					data += instanceStride;
 				}
+
+				// Set vertex and fragment shaders.
+				bgfx::setProgram(program);
+
+				// Set vertex and index buffer.
+				bgfx::setVertexBuffer(vbh);
+				bgfx::setIndexBuffer(ibh);
+
+				// Set instance data buffer.
+				bgfx::setInstanceDataBuffer(idb, numInstances);
+
+				// Bind textures.
+				bgfx::setTexture(0, u_texColor, textureColor);
+				bgfx::setTexture(1, u_texNormal, textureNormal);
+
+				// Set render states.
+				bgfx::setState(0
+					|BGFX_STATE_RGB_WRITE
+					|BGFX_STATE_ALPHA_WRITE
+					|BGFX_STATE_DEPTH_WRITE
+					|BGFX_STATE_DEPTH_TEST_LESS
+					|BGFX_STATE_MSAA
+					);
+
+				// Submit primitive for rendering to view 0.
+				bgfx::submit(0);
 			}
-
-			uint16_t numInstances = (uint16_t)( (data - idb->data)/instanceStride);
-
-			// Set vertex and fragment shaders.
-			bgfx::setProgram(program);
-
-			// Set vertex and index buffer.
-			bgfx::setVertexBuffer(vbh);
-			bgfx::setIndexBuffer(ibh);
-
-			// Set instance data buffer.
-			bgfx::setInstanceDataBuffer(idb, numInstances);
-
-			// Bind textures.
-			bgfx::setTexture(0, u_texColor, textureColor);
-			bgfx::setTexture(1, u_texNormal, textureNormal);
-
-			// Set render states.
-			bgfx::setState(0
-				|BGFX_STATE_RGB_WRITE
-				|BGFX_STATE_ALPHA_WRITE
-				|BGFX_STATE_DEPTH_WRITE
-				|BGFX_STATE_DEPTH_TEST_LESS
-				|BGFX_STATE_MSAA
-				);
-
-			// Submit primitive for rendering to view 0.
-			bgfx::submit(0);
 		}
 
 		// Advance to next frame. Rendering thread will be kicked to 
diff --git a/src/renderer_d3d11.cpp b/src/renderer_d3d11.cpp
index ba614115..449c5fa5 100644
--- a/src/renderer_d3d11.cpp
+++ b/src/renderer_d3d11.cpp
@@ -2672,9 +2672,16 @@ namespace bgfx
 					}
 				}
 
-				if (currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx || programChanged)
+				if (programChanged
+				||  currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx
+				||  currentState.m_instanceDataBuffer.idx != state.m_instanceDataBuffer.idx
+				||  currentState.m_instanceDataOffset != state.m_instanceDataOffset
+				||  currentState.m_instanceDataStride != state.m_instanceDataStride)
 				{
 					currentState.m_vertexBuffer = state.m_vertexBuffer;
+					currentState.m_instanceDataBuffer.idx = state.m_instanceDataBuffer.idx;
+					currentState.m_instanceDataOffset = state.m_instanceDataOffset;
+					currentState.m_instanceDataStride = state.m_instanceDataStride;
 
 					uint16_t handle = state.m_vertexBuffer.idx;
 					if (invalidHandle != handle)
diff --git a/src/renderer_d3d9.cpp b/src/renderer_d3d9.cpp
index 232c481f..1abefe2a 100644
--- a/src/renderer_d3d9.cpp
+++ b/src/renderer_d3d9.cpp
@@ -2679,9 +2679,16 @@ namespace bgfx
 					}
 				}
 
-				if (currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx || programChanged)
+				if (programChanged
+				||  currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx
+				||  currentState.m_instanceDataBuffer.idx != state.m_instanceDataBuffer.idx
+				||  currentState.m_instanceDataOffset != state.m_instanceDataOffset
+				||  currentState.m_instanceDataStride != state.m_instanceDataStride)
 				{
 					currentState.m_vertexBuffer = state.m_vertexBuffer;
+					currentState.m_instanceDataBuffer.idx = state.m_instanceDataBuffer.idx;
+					currentState.m_instanceDataOffset = state.m_instanceDataOffset;
+					currentState.m_instanceDataStride = state.m_instanceDataStride;
 
 					uint16_t handle = state.m_vertexBuffer.idx;
 					if (invalidHandle != handle)
diff --git a/src/renderer_gl.cpp b/src/renderer_gl.cpp
index c5fe1031..2b1adc3c 100644
--- a/src/renderer_gl.cpp
+++ b/src/renderer_gl.cpp
@@ -2754,7 +2754,7 @@ namespace bgfx
 
 	void Context::rendererSubmit()
 	{
-		const GLuint defaultVao = s_renderCtx.m_vaoSupport;
+		const GLuint defaultVao = s_renderCtx.m_vao;
 		if (0 != defaultVao)
 		{
 			GL_CHECK(glBindVertexArray(defaultVao) );
@@ -3274,18 +3274,24 @@ namespace bgfx
 						if (programChanged
 						||  currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx
 						||  currentState.m_indexBuffer.idx != state.m_indexBuffer.idx
-						||  currentState.m_instanceDataBuffer.idx != state.m_instanceDataBuffer.idx)
+						||  currentState.m_instanceDataBuffer.idx != state.m_instanceDataBuffer.idx
+						||  currentState.m_instanceDataOffset != state.m_instanceDataOffset
+						||  currentState.m_instanceDataStride != state.m_instanceDataStride)
 						{
 							bx::HashMurmur2A murmur;
 							murmur.begin();
 							murmur.add(state.m_vertexBuffer.idx);
 							murmur.add(state.m_indexBuffer.idx);
 							murmur.add(state.m_instanceDataBuffer.idx);
+							murmur.add(state.m_instanceDataOffset);
+							murmur.add(state.m_instanceDataStride);
 							murmur.add(programIdx);
 							uint32_t hash = murmur.end();
 
 							currentState.m_vertexBuffer = state.m_vertexBuffer;
 							currentState.m_indexBuffer = state.m_indexBuffer;
+							currentState.m_instanceDataOffset = state.m_instanceDataOffset;
+							currentState.m_instanceDataStride = state.m_instanceDataStride;
 							baseVertex = state.m_startVertex;
 
 							GLuint id = s_renderCtx.m_vaoStateCache.find(hash);
@@ -3351,9 +3357,15 @@ namespace bgfx
 						}
 
 						if (programChanged
-						||  currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx)
+						||  currentState.m_vertexBuffer.idx != state.m_vertexBuffer.idx
+						||  currentState.m_instanceDataBuffer.idx != state.m_instanceDataBuffer.idx
+						||  currentState.m_instanceDataOffset != state.m_instanceDataOffset
+						||  currentState.m_instanceDataStride != state.m_instanceDataStride)
 						{
 							currentState.m_vertexBuffer = state.m_vertexBuffer;
+							currentState.m_instanceDataBuffer.idx = state.m_instanceDataBuffer.idx;
+							currentState.m_instanceDataOffset = state.m_instanceDataOffset;
+							currentState.m_instanceDataStride = state.m_instanceDataStride;
 
 							uint16_t handle = state.m_vertexBuffer.idx;
 							if (invalidHandle != handle)