I am curious if MCIP versions are by default built mainly to exploit parallel computing capabilities, or if they run fine (in terms of computation time) in serial computing configurations as well. I was thinking this scenario will arise when we are limited by parallel computing resources (fixed cpu-hours per person per year).
MCIP is not set up to use parallel computing capabilities. MCIP is largely I/O bound, so my analysis suggests that modifying MCIP for parallel computing would not really help much. I am open to suggestions, if you think otherwise.