Slurm Flags, Partitions, and QoS¶
Slurm allows the use of flags to specify resources needed for a job. Below is a table describing some of the most common Slurm resource flags, followed by tables describing available Summit partitions and Quality of Service (QoS) options.
Slurm Resource Flags¶
Job scripts, the
sbatch command, and the
sinteractive command support many different resource requests in the form of flags. These flags are available to all forms of jobs. To review all possible flags for these commands, please visit the Slurm page on sbatch. Below, we have listed some useful flags to consider when running your job script.
|Allocation||Specify an allocation account||--account=allocation|
|Partition||Specify a partition (see table below)||--partition=partition|
|Sending email||Receive email at beginning or end of job completion||--mail-type=type|
|Email address||Email address to receive the email||--mail-user=user|
|Number of nodes||The number of nodes needed to run the job||--nodes=nodes|
|Number of tasks||The total number of processes needed to run the job||--ntasks=processes|
|Tasks per node||The number of processes you wish to assign to each node||--ntasks-per-node=processes|
|Total memory||The total memory (per node requested) required for the job.
Using --mem does not alter the number of cores allocated
to the job, but you will be charged for the number of cores
corresponding to the proportion of total memory requested.
Units of --mem can be specified with the suffixes: K,M,G,T (default M)
|Quality of service||Specify a QOS (see table below)||--qos=qos|
|Wall time||The max amount of time your job will run for||--time=wall time|
|Job Name||Name your job so you can identify it in the queue||--job-name=jobname|
On Summit, nodes with the same hardware configuration are grouped into partitions. You will need to specify a partition using
--partition in your job script in order for your job to run on the appropriate type of node.
These are the partitions available on Summit.
|Partition||Description||# of nodes||cores/node||RAM/core (GB)||Billing weight||Default/Max Walltime|
|shas||Haswell (default)||380||24||4.84||1||4H, 24H|
|smem 1||High-memory||5||48||42.7||6||4H, 7D|
|sknl||Phi (KNL)||20||68||5.25||0.1||4H, 24H|
|ssky 2||Skylake||5||24||7.68||1||4H, 24H|
|ssky-preemptable 2||Skylake||15||24||7.68||1||4H, 24H|
smempartition is limited to 96 cores (2 entire nodes) across all running smem jobs. For example, you can run one 96-core job or up to two 48-core jobs, four 24-core jobs, ninty-six 1-core jobs, etc. If you need more memory or cores, please contact email@example.com.
2 Skylake nodes are seperated into 2 partitions and consists of 5 public nodes in the
sskypartition and 15 private nodes that can be used by anyone with the
ssky-preemptable. Jobs running on the
ssky-preemptablepartition will be pushed off the node if the condo owner of the node runs a job to their private partition. More information on Skylake nodes can be found here.
In addition to these partitions, Research Computing also provides specialized partitions for interactive and test jobs. These partitions allow quick access to a reserved set of cores provided for testing and interactive use.
|Partition||Description||Max Nodes||Max cores||RAM/core (GB)||Billing weight||Default/Max Walltime|
|shas-testing 3||Haswell||Up to 2||24||4.84||1||0.5H, 0.5H|
|sknl-testing||Phi (KNL)||1||24||5.25||1||0.5H, 0.5H|
shas-testingpartition is limited to 24 cores total. These cores can come from up to 2 nodes but a user is limited to maximum of 24 cores per job.
To run a job longer than 24 hours on the
sknl partitions, use the
More details about each type of node can be found here.
Quality of Service (Summit)¶
On Summit, Quality of Service or QoS is used to constrain or modify the characteristics that a job can have. This could come in the form of specifying a QoS to request for a longer run time or a high priority queue for condo owned nodes. For example, by selecting the
long QoS, a user can place the job in a lower priority queue with a max wall time increased from 24 hours to 7 days.
Normally, this slurm directive does not need to be set for most jobs. Only set a QoS when requesting a long or condo job.
The available QoS’s for Summit are:
|QOS name||Description||Max walltime||Max jobs/user||Node limits||Partition limits||Priority Adjustment|
|long||Longer wall times||7D||200||22/user; 40 total;||shas, sknl, ssky||0|
|condo||Condo purchased nodes only||7D||500||256/user||shas, ssky||Equiv. of 1 day queue wait time|
Couldn’t find what you need? Provide feedback on these docs!