An Efficient Field-Partition Based Code Compression and Its Pipelined Decompression System

A field partition based instruction compression/decompression system for ARM series architecture is proposed. Due to a high degree of repetition in the encoding of the specific fields of the instructions in a program, we could get a statistics of the appearances of each field in all instructions of a program. Depending on the statistics, we partition each instruction into three fields and compress each field using Huffman coding method. Experimental results show that our method is better than others with a 55% of average compression ratio. For decompression, single buffering, double buffering and pipeline techniques have been proposed. However, due to jump penalty, these techniques incur more delays in pipeline or have to stop and fill in the cache buffers. We proposed a pipeline with back-up for flushing technique that incurs no delay and without stopping due to jump. The average performance is increased about 10% to 60%