DirectX Video Acceleration

DXVA2 implementations: native and copy-back

DXVA2 implementations come in two variants: native and copy-back.

With the native implementation, the decoded video stays in GPU memory until it has been displayed. The video decoder must be connected to the video renderer. There are to be no processing filter in between. The video renderer must also support DXVA, which gives you less freedom in the choice of the renderers.

With the copy-back implementation, the decoded video is copied from the GPU memory back to the CPU memory. This implementation doesn't have the limitations mentioned above, acts similar to a normal software decoder, but does require a GPU fast enough in copying data from the GPU memory back to the CPU memory; otherwise the result will be stuttering video playback.

Obviously, if a user has no need for customized processing, then it makes no sense to use the copy-back mode as the copy-back operations increases GPU memory load.

GPUs that should be fast enough are:

AMD: Radeon HD 6xxx and newer
Nvidia: Nvidia GeForce 500 Series and newer
Intel: Intel HD Graphics 2000 and newe