The computing power of GPUs is now accessible to improve the interactivity of 3D graphics. Leveraging this capability requires optimum interoperability between compute (OpenCL) and graphics APIs (OpenGL/DirectX). This talk from AMD demonstrates how to compute and update geometry with OpenCL on APUs, how to update VBO on discrete GPUs, and how to compute physics data on OpenCL and transfer those data to discrete GPUs. Wednesday, 10 August 11:15 AM - 12:15 PM | West Building, Hall C, Booth 801, SIGGRAPH 2011