Differenze

Queste sono le differenze tra la revisione selezionata e la versione attuale della pagina.

--- oph:cluster:accounting [2023/04/06 15:05] – marco.baldi5@unibo.it
+++ oph:cluster:accounting [2023/04/06 15:09] (versione attuale) – marco.baldi5@unibo.it
@@ Linea 7: / Linea 7: @@
 Besides the default account, some users may have **other accounts related to specific projects** with dedicated computational or storage resources on the cluster.
-All accounts have an associated share of the total computing resources of the cluster, defined through an algorithm that has been approved by the OPH Executive Board, that depends on both the number of members of each research sector and the past usage of the cluster resources. New shares are assigned every year by the OPH Executive Board.
+All accounts have an associated **share of the total computing resources** of the cluster, defined through an algorithm that has been approved by the OPH Executive Board, that depends on both the number of staff members of each research sector and the past usage of the cluster resources. New shares are assigned every year by the OPH Executive Board.
-By default, all jobs submitted by the users are **assigned and charged on their default account**, unless a different valid account is specified in the job script with the option:
+By default, all jobs submitted by the users are **assigned and charged on their default account**, unless a different valid account is specified in the job script (see the [[oph:cluster:jobs|Jobs Submission]] section) through the option:
+<WRAP center 40%>
 ''#SBATCH --account="account_name"''
+</WRAP>
-The computational resources already used by a given account (i.e. cumulatively used by all the users associated with that account) will **determine the priority level** of jobs submitted on that account through a **Fair Share algorithm**: the larger the fraction of its share already consumed by an account, the lower its jobs priority will be, so that other accounts with a lower past consumption of their share can more easily access the cluster resources.
+The computational resources already used by a given account (i.e. cumulatively used by all the users associated with that account) will **determine the priority level** of jobs submitted under that account through a **Fair Share algorithm**: the larger the fraction of its share already consumed by an account, the lower its jobs priority will be, so that other accounts with a lower past consumption of their share can more easily access the cluster resources.
 The accounts shares therefore represent a **soft quota** for the use of computational resources: having exhausted the assigned share will not prevent the users of a given account from using the cluster computing resources, but the account priority will be lower than that of other accounts which have not yet consumed their assigned share.