Skip to Main Content
We have implemented the MPEG 1/2 audio Layer-III decoding algorithm using ARM7 and ARM9 based systems. Optimizations are conducted in the algorithm, C-program, and assembly program levels. The assembly program level optimization that employs the block data transfer instructions intensively is very effective for reducing the number of memory accesses. The performance effects of memory architectures, such as the number of internal buses, cache size, and cache schemes are measured both for ARM7 and ARM9 systems. The overhead of clock cycles due to external DRAM accesses is shown about 40% for a typical ARM7 based system, and 19% to 1.6% for typical ARM9 based systems.
Circuits and Systems, 2002. APCCAS '02. 2002 Asia-Pacific Conference on (Volume:1 )
Date of Conference: 2002