Research Projects Supported by HKU's High Performance Computing Facilities |
Researchers: |
Dr Nelson H C Yung and Dr Kwong-keung Leung, Department of Electrical and Electronic Engineering |
Project Title: |
Fast and Efficient Video Coding Based on Communication and Computation Scheduling on Multiprocessors |
Project Description: |
The goal of this project was to investigate possible parallelization methods for video coding under the criteria of efficiency, coding frame rate, coding delay and applicability. To achieve this goal, a philosophy based on the scheduling of data communication and computation on multiprocessors has been adopted and a number of methods and algorithms have been proposed. They include (1) parallelizing the large amount of video data; (2) hiding the communications required between processors when performing coding; (3) generalizing the scheduling of computation tasks; and (4) adaptively schedule tasks in run-time. Simulations on the IBM SP2 multiprocessor indicated that significant speedup and high efficiencies were achieved; and it showed that real-time video coding is indeed possible. |
Project Duration: |
3 years |
Project Significance: |
Video coding techniques, standards and implementations fueled the rapid expansion of applications in multimedia computing, information storage, video phone, video conferencing, remote sensing, medical imaging and many other audiovisual services. Many of these have become an inseparable part of our lives. As video coding is complex by nature that demands intensive computation, our ability to perform coding in real-time determines how well it is being accepted into our daily routines. In this respect, the IBM SP2 enabled us to test the idea/concept. As a result, we were able to apply the algorithms in different applications. One was to collaborate with Texas Instruments’ research center in Tsukuba, Japan, to realize our algorithms on an advanced digital signal processor, with multiple processors. The other was to realize our algorithms on a PDA for real-time video surveillance. Both of them were academic research that resulted into real-life applications with impact to the society. |
Results Achieved: |
We have
achieved real-time coding capability, high speedup of 14.8 over 24
processors on the IBM SP2, and an efficiency of up to 93.8%. |
Remarks on the Use of High Performance Computing Cluster: |
HPC Cluster provided the platform for us to prove our concept. |
Email Addresses: |
nyung@eee.hku.hk (Dr Yung) and kkleungb@graduate.hku.hk (Dr Leung) |