On architectural styles for multimedia signal processors

After presenting several possible multimedia signal processor (MSP) architecture styles, we propose an architecture style which could provide high performance and high flexibilities, and requires less external memory accesses and I/O operations. It is a hierarchical and scalable architecture style which facilitates the hardwaresoftware co-design of MSP circuits and systems. OVERVIEW OF MSP ARCHITECTURES The rapid progress in VLSI technology will soon reach more than 100 million transistors in a chip [13], implying tremendous computation power for many multimedia applications. For example [2, 8, 10, 12], 1. Audio & speech processing: audio compression (G.711, G.722, G.728), surround sound processing, AC-3, etc. 2. Image & video processing: resolution conversion, image enhancement, image restoration, image compression (JBIG, JPEG), video compression (MPEG), etc. 3. Content-based indexing & retrieval: feature extraction (fillet coordination, moment, histogram), pattern recognition, face detection/recognition, fusion of multimodality, etc. 4. 2-D, 3-D, & 4-D graphics: volume rendering, modeling transformation, texture mapping, shading, shadowing, ray-tracing, computer-assisted animation, virtual reality, etc. Some important features of the MSP algorithms can be observed: 1. Intensive computation for highly regular operations There is a great deal of parallelism, especially on common operations such as addition, subtraction, multiplication, etc. 2. Intensive I/O or memory access Multimedia data operands have very frequent and very regular reusability. 3. High control complexity in less computational intensive tasks It may be more efficient and economical to resort to software solution for such tasks. 4. Frequent encounters of small integer operands In MPEG and other pixel-oriented algorithms, the data being operated on are small integers (such as 8-bit or 16-bit), narrower than the existing integer data paths of microprocessors. Subword parallelism must be exploited for higher efficiency. From the above, the following important design issues emerge for MSP: 1. High performance and high flexibility 2. Low cost, low power, and efficient memory usage 3. System level-optimization: system integration or system-on-a-chip 4. Fast design turn-around These objectives may be best achieved by implementing MSP systems in an application specific paradigm supplied by a comprehensive design methodology. As shown in Figure 1, multimedia signal processors can be categorized into (1) ASIC processors, (2) programmable processors (DSP or microprocessor), & (3) custom computing machines.