|
|
Milestones
|
|
|
|
We shall first invest time in isolating and investigate the execution portion in H.264 and
post-processing that can be ported efficiently on GPGPU. After such investigation we shall
be able to come up with a framework design for video decoding.
Thereafter, we shall proceed to the toolset design and development. Specifically, we will
formulate a set of API/SDK to wrap-up the subtasks of video decoding and post-processing,
including but not limited to filtering, deblocking, motion estimation, motion compensation,
DCT, quantization, deblocking, deinterlacing, etc. At the end of this phase, a preliminary
toolset that is functionally correct will be completed.
Once we have completed the toolset, we shall then move on to optimization phase. During
this phase, each module in the toolset will be optimized to run on GPGPU for the best
performance.
Finally, we will integrate the toolset back to the H.264 decoder, and come up with a fully
optimized GPGPU-assisted H.264 decoder. At least two post-processing GPGPU-assisted
algorithms, namely deblocking and deinterlacing, will also be integrated into the final
decoder to demonstrate the power and efficiency of GPGPU coprocessoring approach.
|
|
|
|
|
|
|
