matician09
New member
Hi everyone, I am trying to measure performance of WRF with some kind of benchmark set.
Ideally, to control variability in the system's I/O performance, I would like to run only the minimum required input reads and avoid any output writes. However, when I set
Would anyone be able to offer advice or share good ideas on how to properly measure computational performance?
Ideally, to control variability in the system's I/O performance, I would like to run only the minimum required input reads and avoid any output writes. However, when I set
history_interval as 0,0,0 or 10000,10000,10000 (too big numbers) and run the model, WRF terminates just successfully without performing the main computation. (It appears that at least Domain 1 must be configured for the run to actually execute.)Would anyone be able to offer advice or share good ideas on how to properly measure computational performance?