| Warning |
|---|
By default, if you don’t ask for a specific type of node, the job might be submitted to a GPU node. Be careful about this as it could carry unwanted costs. |
On Computerome 2.0, there are three types of machines:
...
| Code Block |
|---|
| language | powershell |
|---|
| theme | RDark |
|---|
|
$ qsub -W group_list=<group_NAME> -A <group_NAME> -l nodes=1:ppn=40:fatnode,walltime=99:00:00,mem=1200gb -m n <script> |
| Vis_skjul |
|---|
| | Warning |
|---|
By default, if you don’t ask for a specific type of node, the job might be submitted to a GPU node. Be careful about this as it could carry unwanted costs. |
| Info |
|---|
You can add this line to the bottom of your script checkjob -v $PBS_JOBID It will generate something like the following: | Code Block |
|---|
| language | powershell |
|---|
| theme | RDark |
|---|
| Total Requested Tasks: 20
Total Requested Nodes: 1
Req[0] TaskCount: 20 Partition: torque
Dedicated Resources Per Task: PROCS: 1 MEM: 12G
Utilized Resources Per Task: PROCS: 0.37 MEM: 12G SWAP: 2020M
Avg Util Resources Per Task: PROCS: 0.37
Max Util Resources Per Task: PROCS: 0.80 MEM: 12G SWAP: 2020M
Average Utilized Memory: 10761.74 MB
Average Utilized Procs: 8.47 |
To calculate what you should use for the "-l mem=" parameter you have to times the number of tasks with "Max Util Resource Per Task" "MEM:" Here it would bbe 20 * 12 gb = 240gb. |
|
...