NSF_SI2_PI_Meeting2018.pdf

Interactive Parallelization Tool (IPT) assists domain-experts and students in efficiently parallelizing their existing C/C++ applications using any of the following parallel programming models: Message Passing Interface (MPI), OpenMP, and CUDA. For assisting in parallelization, IPT uses its knowledgebase of parallel programming expertise (encapsulated as design templates and rules), and specifications (i.e., what to parallelize and where) as provided by users. It parses the user-specifications and accordingly analyses the input source code to reengineer it for generating a parallel version of the code.IPT can be used for self-­paced learning of different parallel programming paradigms. It helps in understanding the differences in the structure and performance of the parallel code generated for different specifications while using the same serial application.交互式并行化工具（Interactive Parallelization Tool, IPT）可协助领域专家与学生，基于以下任意一种并行编程模型高效完成现有C/C++应用的并行化改造：消息传递接口（Message Passing Interface, MPI）、OpenMP以及统一计算设备架构（Compute Unified Device Architecture, CUDA）。为辅助并行化工作，IPT依托封装为设计模板与规则的并行编程专业知识库，以及用户提供的并行化目标与范围说明（即需并行化的内容及对应位置）。该工具会解析用户提交的说明，并据此对输入的源代码进行分析与重构，以生成对应代码的并行版本。此外，IPT可用于不同并行编程范式的自主学习，能够帮助用户理解在使用同一串行应用程序时，针对不同说明生成的并行代码在结构与性能上的差异。