Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Method for a memory layout for a compressed video data processing system

IP.com Disclosure Number: IPCOM000008007D
Publication Date: 2002-May-10
Document File: 9 page(s) / 92K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for a memory layout for a compressed video data processing system. Benefits include improved performance.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 34% of the total text.

Method for a memory layout for a compressed video data processing system

Disclosed is a method for a memory layout for a compressed video data processing system. Benefits include improved performance.

Background

              Video coding standards, such as those developed by Motion Picture Coding Experts Group (MPEG ), generally imply high computational complexity and affect the total performance of a coding/decoding system (codec).

              Many conventional hardware systems supporting video processing applications implement a multilevel memory model. It is typically comprised of a main memory, and several levels of cache memory that is significantly faster. Ways to increase overall system performance, which may become critical for real-time systems, are to minimize cache memory pollution, optimize cache memory usage, and decrease the number of cache misses.

              A requirement exists for a memory-optimized video-processing scheme because MPEG-like standards process data in a block-based manner. That is, compressed data for a group of operations can be localized in a memory area of predetermined size. Moreover, MPEG-like standards introduce inter-picture dependencies within one video stream. Some data from previous pictures should be stored in a way that complies with the cache requirements and that facilitates future access.

              Block-related memory areas in the scheme should be independent to support parallel processing.

General description

              The disclosed method is a compressed video-data processing system architecture. The method is intended for a block-based compression standard. A raster frame is provided to store display pixel data. An intermediate frame is provided to store coefficients of each block of a video image. All blocks of the video image are placed linearly into the intermediate frame. Descriptor tables are furnished to store some extra information of the video image blocks, and to facilitate access to the intermediate frame. Look-up tables contain references to the descriptor tables’ elements and facilitate access to the descriptor tables.

              An implementation of the disclosed method creates codec internal data structures. The method minimizes cache memory pollution, optimizes cache memory load, and facilitates access to extra picture data. The method includes several buffers for each frame processed, such as previous, current, and next frames for MPEG coding scheme. These buffers are a frame of raster coefficients, a sequence of grouped intermediate coefficients relating to blocks and grouped blocks (macroblocks), and a look-up frame containing extra picture data. Special structures contain references to each block and macroblock, and special fields to map macroblocks to corresponding frame positions.

Advantages

      The disclosed method provides several advantages, including:

§         Cache-optimized video block-based processing that minimizes cache pollution and decreases the number of cache misses.

§         No redundant type-conversion operations

§         Access to extra picture i...