ROMIO: A High-Performance, Portable MPI-IO Implementation

• Developed at Argonne

• Freely available from http://www.mcs.anl.gov/romio

• Runs on most parallel machines and file systems, including Linux clusters

• Works with several MPI-1 implementations

• Included as part of MPICH, SGI MPI, and HP MPI

• Optimized for noncontiguous accesses and collective I/O

Previous slide

Next slide

Back to first slide

View graphic version