I find it a bit strange the way Nvidia has gone with the Dual DMA (Copy Engines), where it is required to have multiple OpenGL context. The solution seems a bit hacky, and I can’t see any reason why this could not be automatically achieved using the asynchronous PBO API from a single OpenGL context by hte driver, which is how AMD seem to plan to support it.
Does Nvidia have any plans on improving the Single/Dual Copy Engine support in OpenGL to automatically enable it from a single OpenGL context?