基于Amdahl定律的分组密码并行处理模型研究

为进行分组密码并行处理的理论研究和定量分析，构建了基于Amdahl定律的分组密码并行处理模型.通过引入分组内数据级并行性、分组内指令级并行性、分组间数据级并行性和分组间指令级并行性等并行性，建立了分组密码的多维度并行处理模型，研究了算法参数和并行参数对分组密码处理性能的影响.结果表明，并行比例大的算法其性能提升空间较大，且开发任一并行性都能够提升密码运算性能；但对于具体的应用场景，需要根据算法结构、工作模式等来进行合理的资源配置和并行性开发；该模型能够为分组密码算法的并行处理架构设计、资源配置和并行性开发等提供理论依据和指导原则. 

Research on Parallel Processing Model for Block Cipher Based on Amdahl's Law

In order to carry out the theoretical research and quantitative analysis of block cipher parallel processing, a block cipher parallel processing model based on Amdahl''s law was established. Introducing data level parallelism in block, instruction level parallelism in block, data level parallelism among blocks and instruction level parallelism among blocks, a multiple dimension parallel processing model was developed to study the effects of algorithm parameters and parallelism parameters on cipher processing performance. The results show that, the algorithms with large parallel ratio possess a larger performance improvement space, and the development of any parallelism can improve the performance. But resource allocation and parallelism development should be carried on according to the algorithm structure and work modes for the specific application scenarios. This model can be taken as the guidance for the parallel processing architecture design, resource allocation and parallelism development of block cipher.