I would like to acknowledge the following people and organizations who have have provided information about MPI implementations or access to MPI implementations: The National Center for Supercomputing Applications (John Townes) provided access to SGI and HP machines, NASA Ames Research Center (Mary Hultquist) provided access to an SGI machine, Lawrence Livermore National Laboratory (John May and Greg Tomaschke) provided access to a Digital Alpha cluster, Argonne National Laboratory (Rusty Lusk) provided access to an IBM SP, Sun Microsystems (Steve Walter) provided access to a beta version of Sun MPI, Rolf Rabenseifner of the University of Stuttgart and Rolf Hempel of NEC provided information about the NEC implementation, Lloyd Llewins of Hughes Aircraft Company provided information about the Mercury RACE implementation, Raja Daoud of HP provided information about the HP and LAM implementations, and Eric Salo of SGI provided information about the SGI implementation.
This work was supported by the Director, Office of Computational and Technology Research, Division of Mathematical, Information, and Computational Sciences of the U.S. Department of Energy under contract number DE-AC03-76SF00098.