TBB as mentioned by Peter has been used by a number of AAA PC games, and they've added functionality to pin (affinitize) certain tasks to a given hardware thread for OpenGL/DX where you need to use the main thread. There are a few good examples of using TBB for games from Intel (full disclosure I used to work there), see http://software.intel.com/en-us/vcsource/samples in particular the tasking update. Intel's GPA Platform Analyzer is handy for multithreading optimizations as well.