This tools aims to complement AMReX's base profilng and profVis in the following aspects:
- provide easy access to profiling results
- analyze per-process time cost along critical call paths
- find where the most expensive functions are called
- detail the high-level expensive functions including what child functions it calls and their costs
This tool is based on AMReX's base profiling. To enable base profiling, one needs to add -DAMReX_BASE_PROFILING
and -DAMReX_TRACE_PROFILING
while building AMReX.
The other dependent python modules (numpy, matpyplotlib, etc) can be easily installed ad-hoc.
The exemplary usages are illusrated in analyze.py
.