mpi4py: Status Update After 12 Years of Development

被引：152

作者：

Dalcin, Lisandro ^{[1
]}

Fang, Yao-Lung L. ^{[2
]}

机构：

[1] King Abdullah Univ Sci & Technol KAUST, Thuwal 23955, Saudi Arabia

[2] Brookhaven Natl Lab, Upton, NY 11973 USA

来源：

COMPUTING IN SCIENCE & ENGINEERING | 2021年 / 23卷 / 04期

关键词：

!text type='Python']Python[!/text; Graphics processing units; Standards; Libraries; Semantics; Programming; Task analysis;

D O I：

10.1109/MCSE.2021.3083216

中图分类号：

TP39 [计算机的应用];

学科分类号：

081203 ; 0835 ;

摘要：

MPI for Python (mpi4py) has evolved to become the most used Python binding for the message passing interface (MPI). We report on various improvements and features that mpi4py gradually accumulated over the past decade, including support up to the MPI-3.1 specification, support for CUDA-aware MPI implementations, and other utilities at the intersection of MPI-based parallel distributed computing and Python application development.

引用

页码：47 / 54

页数：8

共 8 条

[1] Cython: The Best of Both Worlds
Behnel, Stefan
Bradshaw, Robert
Citro, Craig
Dalcin, Lisandro
Seljebotn, Dag Sverre
Smith, Kurt
[J]. COMPUTING IN SCIENCE & ENGINEERING, 2011, 13 (02) : 31 - 39
[2] MPI for Python']Python:: Performance improvements and MPI-2 extensions
Dalcin, Lisandro
Paz, Rodrigo
Storti, Mario
D'Elia, Jorge
[J]. JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING, 2008, 68 (05) : 655 - 662
[3] Fast parallel multidimensional FFT using advanced MPI
Dalcin, Lisandro
Mortensen, Mikael
Keyes, David E.
[J]. JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING, 2019, 128 : 137 - 150
[4] Hammond Jeff R., 2014, 2014 Workshop on Exascale MPI at Supercomputing Conference (ExaMPI). Proceedings, P1, DOI 10.1109/ExaMPI.2014.5
[5] Mohanan A V, 2019, J OPEN RES SOFTWARE, V7, DOI [10.5334/jors.238, DOI 10.5334/JORS.238]
[6] Mortensen M., 2019, J OPEN SOURCE SOFTW, V4, P1340
[7] Mortensen M., 2018, Journal of Open Source Software, V3, P1071
[8] Shafi A., 2021, ARXIV210108878

← 1 →