A new data partitioning scheme for parallel compilation is presented. The scheme is based on partitioning the grammar, which in turn effectively decomposes the language specification into multiple subsets. The scheme requires definition and development of specialized subcompilers. A simple and practical method is suggested for the implementation of such subcompilers using Parser Generators. The proposed scheme is illustrated through an example, using the grammar of Pascal language. A comparison with other methods such as functional decomposition, reveals that the proposed methodology has a larger potential speed-up factor.
提出了一种新的并行编译数据分区方案。该方案基于对语法的划分,从而有效地将语言规范分解为多个子集。该计划需要定义和开发专门的子编译器。建议使用一种简单实用的方法来使用Parser Generators实现此类子编译器。通过使用Pascal语言语法的示例说明了所提出的方案。与功能分解等其他方法的比较表明,该方法具有较大的潜在提速因子。 P>
Dept. of Electrical and Computer Engineering, Syracuse University, Syracuse, NY;
机译:并行PageRank计算的基于站点的分区和重新分区技术
机译:高级并行代码的编译技术
机译:高级并行代码的编译技术
机译:可配置加速器的协同编译中的并行化-主机/加速器分区编译方法
机译:嵌入式数据并行语言的编译技术。
机译:使用基于结构图文法的技术识别所选医学图像的新方面
机译:用于科学代码并行化的高性能Fortran编译技术
机译:统一地下的笔记:基于统一语法形式的论文汇编。