Data Organization and I/O in a Parallel Ocean Circulation Model

We describe an efficient and scalable parallel I/0 strategy for writing out gigabytes of data generated hourly in the ocean model simulations on massively parallel distributed-memory architectures. Working with Modular Ocean Model, using net CIF file system? and implemented on Cray T3E, the strategy speedup I/0 by a factor of 50 in the sequential case. In parallel case, on 8 PEs up to 256 PEs, our implementation writes out most model dynamic fields of about 1GB to a single netCDF file in 65 seconds, independent of the number of processors. The remap-and-write parallel strategy resolves the memory limitation problem and requires minimal collective I/0 capability of the file system. Several critical optimization on memory management and file access are carried out which ensure this scalability and also speedup the numerical simulation due to improved memory management