ALU Augmentation for MPEG-4 Repetitive Padding

In this paper we augment a general purpose ALU with an extra functionality a repetitive padding operation. The proposed solution enables the processor to perform the time exhaustive MPEG-4 padding algorithm in real time. At trivial hardware costs of a few hundred 2x2 AND-OR (or equivalent) logical gates, we achieve an order of magnitude speed-up when compared to software running on a general purpose processor. Our approach allows the MPEG-4 software padding algorithm to run in real time for its most demanding profiles. As an example, a 64-bit pipelined implementation is discussed in details. The approach is general and fits into different architectural concepts and ALU operand widths. More specifically it is shown that 344 extra gates are necessary for a 64-bit implementation and at these expenses a processing speed of more than 7 million macroblocks per second (MB/s) can be achieved. This is an order of magnitude higher than the requirements of the most-demanding MPEG-4 profile levels. Speed and hardware estimations are also reported for 32 and 128-bit ALUs.