CS 6230: Parallel Computing and High Performance Computing

## Assignment 0 0. All code samples and job scripts are available on the course github page. 1. Install compilers and an MPI library (openmpi or mpich) on your laptop. 2. Create accounts and get access to [Tangent][] & [Stampede][]. We will be using these machine for all our assignments and projects. * Read the basic documentation for [Tangent][] as well as [Stampede][] * Compile hello_mpi.c and submit a job on both machines. 3. Write a simple MPI program that does the following, * generate a random integer array, with 100 elements per process, on all processes. hint: initialize the RNG using some function of the rank of the process. * broadcast one value (say the first) from the root (rank 0 or any process really) to all other processes. lets call this the pivot. c int root = 0; int pivot = array[0]; MPI_Bcast( &pivot, 1, MPI_INT, root, MPI_COMM_WORLD ); * on each process compute how many elements are less than the pivot. * Now reduce to obtain the total number of elements (globally) less than the pivot. Reduce to receive the result on the root and print it out. c int localLessThan, globalLessThan; MPI_Reduce(&localLessThan, &globalLessThan, 1, MPI_INT, MPI_SUM, root, MPI_COMM_WORLD); * run on [Tangent][] & [Stampede][] with increasing values of \(p\), i.e., the number of processes. Since we are fixing the number of elements per process to 100, this is a weak scalability experiment. What can you infer about the complexity of MPI_Reduce? Time only the reduction.

[Tangent]: https://wiki.chpc.utah.edu/display/DOCS/Tangent+User+Guide [Stampede]: https://portal.tacc.utexas.edu/user-guides/stampede