DokuWiki

User Tools

Site Tools

Trace:
cluster:173

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
cluster:173 [2018/08/24 18:29]
hmeij07 		cluster:173 [2019/03/25 12:04] (current)
hmeij07
Line 6: Line 6:
 One node ''n37'' has been redone with latest Nvidia CUDA drives during summer 2018.  Please test it out before we decide to redo all of them. It is running CentOS 7.5 and I'm interested to see if programs compiled under 6.x or 5.x break. One node ''n37'' has been redone with latest Nvidia CUDA drives during summer 2018.  Please test it out before we decide to redo all of them. It is running CentOS 7.5 and I'm interested to see if programs compiled under 6.x or 5.x break.
  
-The node is in "inactive" state so your job will go pending. I will force them through.+Use the ''#BSUB -m n37'' statement to target the node. \\ 
 +Update n33-n36 same as n37 (the wrapper is called n37.openmpi.wrapper on all nodes)\\ 
 + --- //[[hmeij@wesleyan.edu|Henk]] 2018/10/08 09:07// \\
  
 Usage is about the same as jobs going to the ''amber128'' queue with two minor changes:  Usage is about the same as jobs going to the ''amber128'' queue with two minor changes: 
Line 36: Line 38:
 #BSUB -q mwgpu #BSUB -q mwgpu
 #BSUB -J "K20 test" #BSUB -J "K20 test"
-#BSUB -m n37+###BSUB -m n37   
 + 
 +#n33-n37 are done and all the same 11Oct2018 
 +# the wrapper is called the same on all host
  
 # cuda 9 & openmpi # cuda 9 & openmpi
Line 102: Line 107:
 #cp -r ~/sharptail/* . #cp -r ~/sharptail/* .
 ## feed the wrapper ## feed the wrapper
-#n37.openmpi.wrapper lmp_mpi-double-double-with-cuda \+#n37.openmpi.wrapper lmp_mpi-double-double-with-pgu \
 #-suffix gpu -var GPUIDX $GPUIDX -in in.colloid -l out.colloid.$LSB_JOBID #-suffix gpu -var GPUIDX $GPUIDX -in in.colloid -l out.colloid.$LSB_JOBID
 ## save results ## save results
Line 110: Line 115:
 </code> </code>
  
-  + 
 +==== ib0 ==== 
 + 
 +We've lost the ability of bringing up interface ''ib0'' when going to 7.5 and the latest kernel. 
 + 
 +Details are described here ... http://www.advancedclustering.com/infinibandomni-path-issue-el-7-5-kernel-update/?sysu=bd584af325e6536411a2bc16ad41b3eb 
 + 
 +Reflecting on this, this is not necessarily that bad.  For GPU compute nodes we do not really need it.  This would also free up 5 infiniband ports on the switch and make the available ports a total of 7.  That could be allocated to new servers we're thinking of buying. 
 + 
 + \\ 
 +**[[cluster:0|Back]]**
cluster/173.1535135375.txt.gz · Last modified: 2018/08/24 18:29 by hmeij07

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
CC Attribution-Share Alike 4.0 International Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki