A STUDY OF I O IN A PARALLEL FINITE-ELEMENT GROUNDWATER TRANSPORT CODE/

A parallel finite element groundwater transport code is used to compar e three different strategies for performing parallel I/O: (1) have a s ingle processor collect data and perform sequential I/O in large block s, (2) use variations of vendor-specific I/O extensions, or (3) use th e extended distributed object network I/O (EDONIO) library. Each proce ssor performs many writes of 1 to 4 kilobytes to reorganize local data in a global shared file. The findings suggest having a single process or collect data and perform large block-contiguous operations may be q uite efficient and portable for up to 32 processor configurations. Thi s approach does not scale well for a larger number of processors becau se the single processor becomes a bottleneck for gathering data. The e ffective application I/O rate observed, which includes times for openi ng and closing files, is only a fraction of the peak device read/write rates. Some form of data redistribution and buffering in remote memor y as performed in EDONIO may yield significant improvements for noncon tiguous data I/O access patterns and short requests. Implementers of p arallel I/O systems may consider some form of buffering as performed i n EDONIO to speed up such I/O requirements.