Compiling collection-oriented languages onto massively parallel computers
Compiling collection-oriented languages onto massively parallel computers
复制标题
将面向集合的语言编译到大规模并行计算机上
DOI：
10.1109/fmpc.1988.47500
复制
发表时间：
1988
期刊：
影响因子：
--
通讯作者：
G. Sabot
中科院分区：
文献类型：
--
作者：
G. Blelloch;G. Sabot
讨论了一些问题和技术，用于编译以收集为导向的语言，以在并行和向量机上使用。描述了用于将嵌套并行性转化为平行平行的技术。讨论了将Paralation LISP子集转换为虚拟机的指令集的编译器。虚拟机的说明是简单的矢量指令，可以进一步转化为广泛的体系结构，包括向量机，单构建并行计算机和多指导平行机器。还实施了另一个转换器，将虚拟机说明转换为连接机的说明（CM-2）。本文中的QuickSort示例是在CM-2上执行的。由于实际的并行硬件通常本质上是平坦的，因此这些技术对高级集合语言的实施者很有用。这些语言以一种自然的方式表达并行性，而这种方式起初似乎无法得到大量并行机器的支持。这些技术从嵌套集合和操作的便利和力量到CM-2等机器的速度和简单性形成了桥梁。<< etx >>
Some issues and techniques are discussed for compiling collection-oriented languages for use on parallel and vector machines. Techniques for translating nested parallelism into flat parallelism are described. A compiler that translates a subset of Paralation Lisp into the instruction set of a virtual machine is discussed. The instructions of the virtual machine are simple vector instructions which could be further translated to a broad class of architectures including vector machines, single-instruction parallel machines, and multiple-instruction parallel machines. Another translator is also implemented that translates the virtual machine instructions into instructions for the Connection Machine (CM-2). The quicksort example in this paper is executed on the CM-2. Since actual parallel hardware is usually flat in nature, these techniques are useful to implementors of high-level collection-oriented languages. These languages express parallelism in a natural way that at first does not appear to be supportable by massively parallel machines. These techniques form a bridge from the convenience and power of nested collections and operations to the speed and simplicity of machines like the CM-2.<<ETX>>