On Cirrus , we get a performance fail when comparing slurm energy and cray power counters since the image update today.
The difference between cray power counters and slurm energy seems to be about 50-200J , so small relative to the energy from the slurm job.
This does not seem to be caused from a bug in the script , but inaccurancy in how the energy is reported on the system
On Cirrus , we get a performance fail when comparing slurm energy and cray power counters since the image update today.
The difference between cray power counters and slurm energy seems to be about 50-200J , so small relative to the energy from the slurm job.
This does not seem to be caused from a bug in the script , but inaccurancy in how the energy is reported on the system