The tool is capable of running and reporting the following performance tests:
Strong Scaling tests: Reports how the model scales by incrementing processors count while maintaining the same problem size.
Compilation Efficiency tests: Compares different build types (e.g. different compilers, libraries…) scalability.
Weak Scaling tests: Reports the model speed scaling both the problem size and efficiency.
Model Tracing: Reports more complex performance metrics (PAPI counters, MPI profile, User Functions…) by tracing the execution of the model.
HW Benchmarks: Tests and reports HW benchmarks of the current HPC platform to discard issues in the platform.