Abstract: Techniques related to parallel partitioning and coding mode selection for improved video coding throughput are discussed. Such techniques include performing parallel partitioning and coding mode selection for a lower-right coding unit of a first largest coding unit and an upper-left coding unit of a second largest coding unit to the right of the first largest coding unit and, immediately subsequent thereto, performing parallel partitioning and coding mode selection for a lower-left coding unit and an upper-right coding unit of the second largest coding unit.

Abstract: Independent workloads may be grouped together into a single super workload. This super workload is dispatched to a single context hardware system that does not run an operating system. This effectively creates a multi-context system out of a single context hardware processor.

Abstract: Techniques related to parallel partitioning and coding mode selection for improved video coding throughput are discussed. Such techniques include performing parallel partitioning and coding mode selection for a lower-right coding unit of a first largest coding unit and an upper-left coding unit of a second largest coding unit to the right of the first largest coding unit and, immediately subsequent thereto, performing parallel partitioning and coding mode selection for a lower-left coding unit and an upper-right coding unit of the second largest coding unit.

Abstract: In some embodiments, independent workloads may be grouped together into a single super workload. This super workload is dispatched to a single context hardware system that does not run an operating system. This effectively creates a multi-context system out of a single context hardware processor.

Abstract: A video compression method of estimating a motion vector representative of a difference between a reference frame and a current frame is described. Typically, both the reference frame and the current frame include background and foreground data. In the method a block of the current frame is selected for analysis, and a determination is made whether the block selected is a foreground block or a background block. If the block is a foreground block, then it is compared with predetermined search points in the reference frame. On the other hand, if the block is a background block, it is compared with predetermined search points in only a portion of the reference frame.

Abstract: A video compression technique is provided which reduces motion estimation computations. A digital signal processing system employs external memory. Detection speed is improved by loading a succession of refined search windows are loaded on-chip. By so doing, the search involves fewer accesses to external memory and so completes in a shorter amount of time.