I had recently installed intel parallel studio xe 2018u4.
I have gathered some data for the purpose of application profiling on intel 8280 CPU using the intel APS tool (shipped with 2018u4), but i did not enable the system wide monitoring while using APS binary. For Example, I did not -
1) set the /proc/sys/kernel/perf_event_paranoid value to 0 and,
2) install the VTune Profiler/sampling drivers .
I skipped aforementioned steps as it was listed as optional, and i am not too worried about collection overhead.
I have gathered data using following commands -
a) mpiexec.hydra -genvall -np $SLURM_NPROCS -ppn $SLURM_NTASKS_PER_NODE aps -c mpi ./wrf.exe
b) mpiexec.hydra -genvall -np $SLURM_NPROCS -ppn $SLURM_NTASKS_PER_NODE aps ./wrf.exe
The data gathered using aforementioned commands is valid for the purpose of analysis? as i had noticed sometimes back that there was a significant difference in the performance metrics data with b) command ( Linux Perf vs PMU analysis )
I hope that the mpi data gathered using "aps -c mpi" should be good (not require drivers)? - as per this document it seems the drivers help only in the data collection metrics such as HPC Performance Characterization, Memory Access, and Microarchitecture Exploration.
yes, in case your perf_event_paranoid lets to gather PMU metrics at least in per process mode then it is valid for analysis.
And yes "aps -c mpi" will gather only MPI related statistics and no PMU collector will be launched in this mode at all.
also please switch to the latest and greatest version, currently this is VTune 2020 Gold.
>>as i had noticed sometimes back that there was a significant difference in the performance metrics data with b) command ( Linux Perf vs PMU analysis )
Could you please write a little more details what exactly was different?