Team Member:  Mark Lodermeier

Title: Sum of Absolute Differences Hardware Accelerator For Motion
Estimation Applications

Abstract: The Sum of Absolute Differences (SAD) function is often the
most time consuming operation for motion estimation and compensation. 
Therefore, separate hardware accelerators are often employed to achieve
real-time performance for multimedia applications.  SAD hardware
accelerators exploit the fact that block search algorithms perform
multiple calculations on the same set of data, creating opportunity for
parallelism.  With the advent of the newest international video coding
standard, MPEG-4 part 10 – AVC, new techniques need to be employed in
these accelerators, such as variable block size support for blocks as
small as 4x4, as large as 16x16, and the corresponding permutations.  An
in-depth look into a possible solution for providing variable block size
support for motion estimation has been performed.