We would like to inform you that starting from Thursday, March 9th, 2023, we are changing the method of calculating fairshare. We are adding a new coefficient called "spec", which takes into account the speed of the computing node on which your job is running.
Until now, "usage fairshare" was calculated as
usage = used_walltime*PE , where "PE" represents processor equivalents expressing how many resources (ncpus, mem, scratch, gpu...) the user allocated on the machine.
From now on it will be calculated as
usage = spec*used_walltime*PE , where "spec" denotes the standard specification of the main node (spec per cpu) on which job is running. This coefficient takes values from 3 to 10.
We hope that this change will allow you to use our computing resources even more efficiently. If you have any questions, please do not hesitate to contact us.