User Tools

Site Tools


memory_reservation

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
memory_reservation [2017/07/09 15:34]
meesters
memory_reservation [2018/06/27 10:12]
henkela
Line 10: Line 10:
  
 <code shell> <code shell>
-$ JobID%20,​ReqMem,​MaxRSS,​AveRSS,​Elapsed,​CPUTime -j JOBID+sacct -o JobID%20,​ReqMem,​MaxRSS,​AveRSS,​Elapsed,​CPUTime -j <JOBID>
 </​code>​ </​code>​
 where JOBID is the one you’re interested in. This sample command gives the output for all JobSteps and compares used CPU time with the actual elapsed time - sometimes useful to get performance hints. If your job completed long in the past you may have to tell ''​sacct''​ to look further back in time by adding a start time with -S YYYY-MM-DD. Note that for parallel jobs spanning multiple nodes, this is the maximum memory used on any one node; if you’re not setting an even distribution of tasks per node (e.g. with ''​%%--%%ntasks-per-node''​),​ the same job could have very different values when run at different times. where JOBID is the one you’re interested in. This sample command gives the output for all JobSteps and compares used CPU time with the actual elapsed time - sometimes useful to get performance hints. If your job completed long in the past you may have to tell ''​sacct''​ to look further back in time by adding a start time with -S YYYY-MM-DD. Note that for parallel jobs spanning multiple nodes, this is the maximum memory used on any one node; if you’re not setting an even distribution of tasks per node (e.g. with ''​%%--%%ntasks-per-node''​),​ the same job could have very different values when run at different times.
Line 24: Line 24:
 The CPU time divided by the number of used CPUs should more or less equal elapsed run time. Otherwise, this is an indication for poor parallelisation. The CPU time divided by the number of used CPUs should more or less equal elapsed run time. Otherwise, this is an indication for poor parallelisation.
  
 +<WRAP center round info 80%>
 +On **Mogon 2** we use a JobSubmit-Plugin to manage memory reservation in case the user didn't speficy one. If the job is going to the broadwell nodes the default memory per node is 57.000MB. On the other hand, if the job is going to the skylake nodes the default memory per node will be set to 88.500MB.
 +For the bigmem-partition the process runs vice versa. If you specify memory per node > 1TB without specifying constrain broadwell or skylake you'll get skylake since that's the only nodes that support up to 1,5TB. For anything < 1TB you'll get broadwell if you don't specify anything different.
 +</​WRAP>​
memory_reservation.txt · Last modified: 2018/06/27 10:12 by henkela