Verfahren und Vorrichtung für die effiziente Adressierung des DRAMs eines Videodekompressionsprozessors

摘要

Pixel data is stored and subsequently read from a random access memory (22) of a video decompression processor (20) in a manner that reduces the number of times different rows of the RAM (22) must be addressed in order to retrieve portions of the pixel data therefrom. Pixel data from a video frame (70) is stored in the RAM as a plurality of pages. Each page substantially fills a different row of the RAM (22) and corresponds to a different section of the video frame (70). A motion vector (MV) is decoded to determine the location of a prediction area (90) within the video frame (70). In the event that the prediction area (90) encompasses more than one of the pages of the video frame (70), the pixel data is retrieved one page at a time, minimizing the number of row changes required when addressing the RAM (22) to retrieve the data. <MATH>