Introduction


Up: Contents Next: Linking and running programs

MPICH is a freely available implementation of the MPI standard that runs on a wide variety of systems. The details of the MPICH implementation are described in [6]; related papers include [3] and [5]. This document assumes that MPICH has already been installed; if not, you should first read Installation Guide to MPICH, a Portable Implementation of MPI [4]. For concreteness, this document assumes that the MPICH implementation is installed into /usr/local/mpich and that you have added /usr/local/mpich/bin and /usr/local/mpich/build/<arch>/<device>/bin to your path. If MPICH is installed somewhere else, you should make the appropriate changes.

In addition, you need to know what systems MPICH has been built and installed for. You need to know the architecture and the device. The architecture indicates the kind of processor and/or OS; examples are sun4 and intelnx. The device indicates how MPICH performs communication between processes and how processes are created; examples are ch_nx (for the Intel Paragon, an MPP), ch_p4 (for workstation clusters), and globus (for ``computational grids'' that include both MPPs and workstations, and that may span multiple sites). The libraries are provided in the directory /usr/local/mpich/build/<architecture>/<device>/lib, and the special commands for each architecture/device pair are provided in the directory /usr/local/mpich/build/<architecture>/<device>/bin. For example, the library directory for the sun4 architecture and ch_p4 device is in /usr/local/mpich/build/sun4/ch_p4/lib and the special commands are located in /usr/local/mpich/build/sun4/ch_p4/bin. This directory should also be in your path (or you should use the full path for some commands; we'll indicate which). (Note that sun4 is MPICH's name for SunOS before Solaris; solaris is used for Solaris also known as SunOS version 5.)

This approach makes it easy to have MPICH available for several different parallel machines. For example, you might have a workstation cluster version and a massively parallel version.



Up: Contents Next: Linking and running programs