Methods, apparatuses and systems directed to a video codec that facilitates parallel processing operations is implemented on a graphics processing unit or other hardware device. The codec encodes video frames by dividing a frame into macroblocks and processing the macroblocks in parallel either via plane fitting operations or by motion estimation adjustments. A gathering operation writes the data of each macroblock onto an output stream in a parallel processing operation to accelerate writing the output from the memory of the graphics processing unit to a main memory of a hardware system.