基于OpenMP的文件压缩与解压的并行设计模型

File compression and decompression parallel design model based on OpenMP

在多核环境下,对文件压缩与解压并行算法进行研究,提出一种基于OpenMP的文件压缩处理并行设计模型。该模型由查找热点代码、并行化分析、并行建模、实现、调试等步骤组成。以动态哈夫曼算法为研究算法,将多核压缩处理并行设计模型应用到文件压缩与解压中。并在文件并行处理过程中,与数据分解法相结合对数据文件进行分割,将分解后的数据由主线程分给多个处理器上的多个子线程来并行处理,以此提高多核处理器的利用率并提高文件压缩效率。最后通过实验模拟验证模型以及算法性能。研究结果表明：在八核处理器下通过对文本文件、图像文件和音频文件等多种不同类型文件进行压缩解压试验,验证了动态Huffman并行算法与串行算法相比其加速比可以达到1.5~8.0倍,性能也得到很大提高。

The parallel algorithm was studied for the compression and decompression of file under multi-core environment. The file compression processing parallel design model was proposed based on OpenMP, and the procedures of the model include finding the hot spot, parallel analsis, modeling, realizing and debugging. Taking the dynamic Huffman algorithm as the research algorithm, the multi-core compression processing parallel design model was applied in the compression and decompression of file. To improve the utilization of multi-core processor rate and speed up the file compression efficiency, the method of data decomposition was used to decompose data for parallel processing when the parallel algorithm deals with file compression and decompression. Finally, experimental evaluation was performed to demonstrate the efficiency. The results show that conducting compression and decompression test through different files such as word documents, image files and audio files in the octa-core processor, the speedup of the dynamic Huffman parallel algorithm is 1.5 to 8.0 times higher than that of the serial algorithm, and the performance is greatly improved.