HOME > PARALLEL HDF > PHDF5
1. Design Overview
1.1. Function requirements
- An API to support parallel file access for HDF5 files in a message passing environment.
- Fast parallel I/O to large datasets through standard parallel I/O interface.
- Processes are required to do collective API calls only when structural changes are needed for the HDF5 file.
- Each process may do independent I/O requests to different datasets in the same or different HDF5 files.
- Supports collective I/O requests for datasets (to be included in next version).
- Minimize deviation from HDF5 interface.
1.2. System requirements
- C and Fortran (F9x) language interface.
- Use Message Passing Interface (MPI) for interprocess communication.
- Use MPI-IO calls for parallel file accesses.
- Parallel File system.
2. Features supported
2.1. Datasets
- Fixed dimension sized datasets
- Extendible dimension sized datasets
- Chunked storage datasets
- Compression support: read only, no write.
2.2. Datatypes
- Integer, Float, String classes
- Variable sized type support: read only, no write.
2.3. I/O mode
- Independent read or write
- Collective read or write (collective as defined in MPI)
2.4. Limits
- Chunked storage (including extendible dimension sized datasets) does not support writing to overlapping chunks. That is process m and process n do not writing to the same chunk at the same time. If that happens, the result is undetermined.
- No write for compressed datasets.
- No write for variable length datatypes.
- - Last modified:August 15th 2007