Aggressive Optimization

For science experiments and production runs (such as SMAP L4_SM), aggressive optimization is recommended.

Aggressive optimization results in faster execution of GEOSldas.x, but the simulation results are not zero-diff vs. regular optimization or across different MPI layouts.

To obtain a build with aggressive optimization, use parallel_build.csh -aggressive, which will build in build-Aggressive/ and install in install-Aggressive/.

Based on a 1-year, global Catchment model simulation and a 6-month, global L4_SM-like data assimilation experiment conducted in February 2021, aggressive optimization executes 20-25% faster than regular ("release") optimization. The differences between aggressive and regular optimization are larger than roundoff differences but are not scientifically meaningful. Larger differences were found for the data assimilation experiment.