Executive Summary

The high compression provided by H.264 comes at the expense of high computational complexity and requires a lot of processing power. To cater for this kind of processing power, the latest trends are based on multicore media processor solutions for embedded devices. Multicore media processors are based on the idea of scalability of processing cores rather than the clock speed, thus providing high processing power by avoiding the limiting factor of power consumption at high clock speed.

The task being tackled here is the analysis, design and porting of the H.264 video decoder to run on heterogeneous multi-core processors. The focus will be to develop software architecture to address the issues of parallel processing, synchronization and interprocess communication to meet the needs of different heterogeneous multicore architectures. The proposed architecture will be ported to TI’s DM 6467 heterogeneous multicore chip for real time H.264 decoding. This real time solution will provide 30 fps for baseline and main profile standards up to level 3.0, as specified by the ITU-T.

With digital multimedia becoming a part of our daily lives, the proposed solution will be a step towards facilitating product development in this field and existing multimedia OEMs will greatly benefit from this development.

Key benefits of this project are given below:

• Companies all over the world developing multimedia consumer products come under the category of direct customers / beneficiaries of the project. These multimedia products range from handheld multimedia devices, IPTV and video playback devices etc. These solutions are based on different multimedia processors. The companies will be able to use the C implementation and port it to any processor or companies using TI’s Davinci as the processor in their product will benefit from the outcome of the project and will be able to buy off the shelf H.264 decoding solution and speed up their time to market. The cost effectiveness of the solution can be an important factor, especially when the products are developed in large numbers.

• Researchers and companies working on the multimedia technologies can use the outcome of the project to evaluate different heterogeneous multi-core processors as a candidate for their solution and H.264 as a targeted video decoder.

• Researchers and companies working on multi-core multimedia processors will be able to get helpful insight into the bottlenecks that arise while porting data and computational intensive video decoders to multiple cores and will be able to counter these bottlenecks based on the outcomes of this project.

• The techniques developed will be applicable to other decoders like MPEG-4, VC-1 and AVS and will allow the development teams all around the world to take advantage of it and use these techniques to expedite their real time implementations and hence cut the time to market their solutions.