Fiona J. L. Reid
EPCC, The University of Edinburgh
Two different versions of NEMO (2.3 and 3.0) have been compiled and tested on HECToR. The performance of these versions has been evaluated and an optimal processor count suggested. The NEMO code is found to scale up to 1024 processors with the best performance in terms of AU usage being obtained between 128 and 256 processors. Square grids are found to give the best performance and where these cannot be used, choosing the grid dimensions such that jpni < jpnj is found to give the best performance. The removal of land only cells reduces the number of AU's by up to 25% and also gives a small reduction in the total runtime.
NetCDF 4.0, HDF5 1.8.1, zlib 1.2.3 and szip have been installed and tested. NetCDF 4.0 is found to give considerable reduction to both the amount of I/O and time taken in I/O when using the NOCSCOMBINE tool. The version of netCDF 4.0 installed under this dCSE project is 8-20% faster than the version installed centrally (via modules) on the system. NEMO has been adapted to use netCDF 4.0 for its main output files resulting in a reduction in file size of up to 3.55 times relative to the original code.
Nested models have also been investigated. The BASIC nested model has been compiled and tested and problems with the time step interval identified and rectified. The performance of the BASIC model has been investigated and an optimal processor count (in terms of AU usage) found to be 32.