Difference between revisions of "Resource list"

From Nuclear Physics Group Documentation Pages
Jump to navigationJump to search
 
Line 1: Line 1:
Several resources can be requested when a job is submitted. A full list of resources can be found [http://www.clusterresources.com/torquedocs21/2.1jobsubmission.shtml#resources here].  
+
Several resources can be requested when a job is submitted. A full explanation of the resource list can be found [http://www.clusterresources.com/torquedocs21/2.1jobsubmission.shtml#resources here]. This page includes a list of features which have been tested on Endeavour.
  
By default PBS sends jobs to the last available node. Once resources on the first node are exceeded jobs will carry over to lower nodes.
+
;mem=[size] : Specifies the maximum mount of memory used by the job. [size] is written as an integer followed by a suffix specifying the size multiplier (b, kb, mb, gb). For example, mem=200mb would specify a maximum 200mb of memory for the job.
  
You can specify nodes as a number, or by naming nodes individually by hostname.
+
;nodes : Specifies the nodes used for the job. By default PBS sends jobs to the last available node. Once resources on the first node are exceeded jobs will carry over to lower numbered nodes. You can specify nodes as a number, or by naming nodes individually by hostname. Specifying nodes by number defaults to using one processor per node. Using ppn=[1-8] allows you to specify a number of processors per node. <br/><br/> You can also specify nodes individually by host name. The syntax would be nodes=node2+node3+node4. Nodes requested in this way must be requested in sequence, for example specifying nodes=node3+node7+node9 would not work.
Specifying nodes by number (e.g. "qsub -l nodes=8") seems to behave as if it
 
assumes nodes are the same as processor cores. Specifying 9 nodes in this case
 
would use all 8 processors on node 9, and one processor on node 8. This seems
 
to work up to 27 processors, after which it returns the error:
 
  
"qsub: Job exceeds resource limits MSG=cannot locate feasible nodes"
+
:'''Examples: '''
  
I'm still working on figuring out why this is the case.
+
:Selecting 8 nodes with 1 processor per node:
  
Individually naming nodes will allow you to select specific nodes, and to
+
qsub -l nodes=8 script.sh
assign jobs to all nodes. The syntax for this is "qsub -l
+
 
nodes=node4+node5+node6". You can specify as many nodes as you want in this
+
:Selecting 8 nodes with 4 processors per node:
way, but it doesn't seem to work if you choose nodes out of sequence (3,5, and
+
 
9, for example). It seems that specifying nodes this way defaults to using one
+
qsub -l nodes=8:ppn=4 script.sh
processor per node. You can, however, specify a number of processors to use on
+
 
each node. The syntax is "qsub -l nodes=node2:ppn=8+node3:ppn=2" (which would
+
:Selecting nodes 2, 3 and 4 with 8 processors per node:
specify 8 processors on node2 and two on node3).
+
 
 +
qsub -l nodes=node2+node3+node4:ppn=8
 +
 
 +
:Selecting 8 processors on node 2 and three processors on node 3:
 +
 
 +
qsub -l nodes=node2:ppn=8+node3:ppn=3 script.sh

Revision as of 11:53, 14 September 2009

Several resources can be requested when a job is submitted. A full explanation of the resource list can be found here. This page includes a list of features which have been tested on Endeavour.

mem=[size]
Specifies the maximum mount of memory used by the job. [size] is written as an integer followed by a suffix specifying the size multiplier (b, kb, mb, gb). For example, mem=200mb would specify a maximum 200mb of memory for the job.
nodes
Specifies the nodes used for the job. By default PBS sends jobs to the last available node. Once resources on the first node are exceeded jobs will carry over to lower numbered nodes. You can specify nodes as a number, or by naming nodes individually by hostname. Specifying nodes by number defaults to using one processor per node. Using ppn=[1-8] allows you to specify a number of processors per node.

You can also specify nodes individually by host name. The syntax would be nodes=node2+node3+node4. Nodes requested in this way must be requested in sequence, for example specifying nodes=node3+node7+node9 would not work.
Examples:
Selecting 8 nodes with 1 processor per node:
qsub -l nodes=8 script.sh
Selecting 8 nodes with 4 processors per node:
qsub -l nodes=8:ppn=4 script.sh
Selecting nodes 2, 3 and 4 with 8 processors per node:
qsub -l nodes=node2+node3+node4:ppn=8
Selecting 8 processors on node 2 and three processors on node 3:
qsub -l nodes=node2:ppn=8+node3:ppn=3 script.sh