TACO-Template Based Collections for Distributed Computing Platforms

摘要

Collective operations on distributed data sets promote a high-level data-parallel programming style that eases many aspects of parallel programming on distributed computing platforms. In this paper we describe the design and im-plementation of TACO, a template library that extends the Multiple Threads Template Library with Higher-order oper-ations on distributed object sets by means of reusable topol-ogy classes and C++function templates.