MVAPICH2

A Highly Scalable MPI Implementation for High-End Computer Systems

Introduction

MVAPICH2 is an implementation of MPI (Message Passing Interface) for use with several different network architectures including InfiniBand, OmniPath, Ethernet and RoCE. It is designed to be highly scalable and very fault tolerant which makes it ideal for use with high-end computer systems such as those found at RCC.

Using MVAPICH2 on RCC Resources

There are a few different varieties of MVAPICH2 available on RCC resources. The GNU MVAPICH2 implementation is available as is the Intel one and one for PGI. To access these, a module needs to be loaded for each. For specific usage information, please refer to the main website.

For GNU MVAPICH2

module load gnu-mvapich2

For Intel MVAPICH2

module load intel-mvapich2

For PGI MVAPICH2

module load pgi-mvapich2