MPI (General)

  1. Performance Penalty for MPI Communication
  2. Fabric-Based Collective Offload Solution
  3. General run-time tuning for Open MPI 1.4 and later (Part1)


  1. Compiling MVAPICH2-1.9 with Intel and CUDA


  1. Building OpenMPI with Intel Compilers
  2. Compiling OpenMPI 1.6.5 with Intel 12.1.5 on CentOS 6
  3. Compiling OpenMPI 1.7.2 with CUDA and Intel Compilers 13
  4. Registering sufficent memory for OpenIB when using Mellanox HCA
  5. Building OpenMPI Libraries for 64-bit integers
  6. Compiling OpenMPI-3.1.6 with GCC-6.5