collectives.f  -- demonstrates basic usages of collective MPI functions
gather.f       -- both gather.f and scatter.f demonstrate finer points
scatter.f      -- of MPI_Gather and MPI_Scatter; like padding
transpose_scatter.f  -- demonstrates the use of MPI_Scatter for matrix
                        transposition
transpose_alltoall.f -- demonstrates the use of MPI_Alltoall for matrix
                        transposition
To compile the above programs, use
Lee% make -f make.bgl              <== bluegene
Twister% make -f make.aix          <== pSeries
Cootie% make -f make.ch_gm.intel   <== Linux Cluster, Intel compiler
Cootie% make -f make.ch_gm.gnu     <== Linux Cluster, GNU compiler
Katana% make -f make.katana.pgi    <== Katana Cluster, Portland Group compiler
(System default is PGI compilers)
Katana% make -f make.katana.gnu    <== Katana Cluster, GNU compiler
(To use the GNU compilers, setenv MPI_COMPILER gnu)

To submit job to the BG, either prepare your own batch script or
use the commandline script bglsub (see scv.bu.edu for detail) as follows:

Lee % bglsub 4 $PWD $PWD/collectives

Last updated on: June 24, 2008

