OMB-Py
收藏资源简介:
OMB-Py是由俄亥俄州立大学计算机科学与工程系开发的一个用于评估Python中MPI库在HPC系统上性能的微基准测试套件。该数据集旨在填补目前缺乏针对mpi4py和Python MPI代码在现代HPC系统上通信性能评估的空白。OMB-Py包含多种点对点和集体通信基准测试,支持包括NumPy、CuPy、Numba和PyCUDA在内的流行Python库。数据集的创建旨在帮助Python HPC社区最大化效率,通过提供必要的工具来表征他们正在工作的系统的性能,特别是在MPI通信性能方面。
OMB-Py is a microbenchmark suite developed by the Department of Computer Science and Engineering at The Ohio State University, designed to evaluate the performance of MPI libraries in Python on high-performance computing (HPC) systems. This dataset aims to fill the current gap in the evaluation of communication performance for mpi4py and Python MPI code on modern HPC systems. OMB-Py includes a variety of point-to-point and collective communication benchmarks, and supports popular Python libraries such as NumPy, CuPy, Numba, and PyCUDA. It was created to help the Python HPC community maximize efficiency by providing essential tools to characterize the performance of the systems they utilize, particularly regarding MPI communication performance.



