Open
Description
What would you like to be added?
This issue is based on @rootfs @marceloamaral 's idea during the community meeting.
As an extension from the user-provided spec file in sustainable-computing-io/kepler-model-server#231 (comment), we should further provide a default spec mapping between processor model, so that in runtime, for the case that Kepler on VM can get the processor model name from lscpu.
The profiles are created beforehand and tracked in a repo (could be separate repo). These profiles are used by kepler or kepler-operator to create configMaps.
Kepler probes the CPU model and vCores on the fly and finds the best matches from the default VM profiles to use the idle power models.
Why is this needed?
To provide machine spec for estimator to find the right power model.