wiki.chpc.ac.za

User Tools

Site Tools

You are here: start » howto » quantum_espresso_checkpointing_restarting
Trace: • pythia8 • pbsdsh • feko • namd • vmd • openfoam • wrf
howto:quantum_espresso_checkpointing_restarting

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
howto:quantum_espresso_checkpointing_restarting [2021/10/05 12:26]
alopis 		howto:quantum_espresso_checkpointing_restarting [2021/12/09 16:42] (current)
Line 3: Line 3:
 Most codes permit a checkpoint file to be created so that one's job can be continued from the point shortly before it failed. Since QE 5.1 restarting from an arbitrary point of the code is no more supported.  Instead it is possible to write out only one checkpoint file at a user specified time. Most codes permit a checkpoint file to be created so that one's job can be continued from the point shortly before it failed. Since QE 5.1 restarting from an arbitrary point of the code is no more supported.  Instead it is possible to write out only one checkpoint file at a user specified time.
  
-Please see and look for restart_mode and max_seconds in the QE documentation.+Please see and look for restart_mode and max_seconds in the QE documentation.\\ 
 https://www.quantum-espresso.org/Doc/INPUT_PW.html#idm60 https://www.quantum-espresso.org/Doc/INPUT_PW.html#idm60
  
-If your QE job is running on one node, the //smp// queue, then the maximum walltime is 96 hours, while for 2 nodes or more on the //normal// queue the maximum walltime is 48 hours.+Please specify your outdir which is where the scratch files will be written.\\ 
 +outdir='  '
  
-If you believe your job will require a shorter time than the maximum walltimes, then it is important to use this fact. You also may believe that you job may not complete in the 48 or 96 hours. You can then make a decision on what //max_seconds// value to use in your QE input file. For example:+If your QE job is running on one node, the //smp// queue, then the maximum walltime is 96 hours, while for 2 nodes or more on the //normal// queue the maximum walltime is 48 hours.
  
-1. You believe your job will complete within 20 hours for example, or that around 20 hours is a vital point in the calculation where you believe restarting from is important, or that you've seen your job fail previously after 20 hours:+If you believe your job will require a shorter time than the maximum walltimes, then it is important to choose a shorter PBS walltime. You also may believe that you job may not complete in the 48 or 96 hours. You can then make a decision on what //max_seconds// value to use in your QE input file. For example:\\  
 +\\ 
  
-Set your PBS job walltime greater than 20 hours (21 hours for example), and choose max_seconds equivalent to 20 hours.+**1.** You believe your job will complete within 20 hours for example, or that around 20 hours is a vital point in the calculation where you believe restarting from is important, or that you've seen your job fail previously after 20 hours:
  
-20 hours = 20 x 60 x 60 = 72 000 seconds +Set your PBS job walltime greater than 20 hours (23 hours for example), and choose max_seconds equivalent to 20 hours.
-max_seconds=72000 +
-restart_mode ='restart'+
  
-2. Your job won't complete in 48 hours on //normal// queue, then choose a time less than 48 hours such as 47 hours:+20 hours = 20 x 60 x 60 = 72 000 seconds\\  
 +\\ 
 +Initial first job:\\ 
 +max_seconds=72000\\  
 +restart_mode ='from_scratch'\\  
 +\\  
 +Subsequent jobs (2nd, 3rd etc):\\ 
 +max_seconds=72000\\  
 +restart_mode ='restart'\\  
 +\\ 
  
-47 hours = 47 x 60 x 60 = 169 200 seconds 
-max_seconds=169200 
-restart_mode ='restart' 
  
-2. Your job won't complete in 96 hours on //smp// queue, then choose a time less than 96 hours such as 95 hours:+**2.** Your job may not complete in the maximum PBS walltime of 48 hours on the //normal// queue, so then choose max_seconds equivalent to less than 48 hours, such as 45 hours:\\ 
  
-95 hours = 95 x 60 x 60 = 342 000 seconds +45 hours = 45 x 60 x 60 = 162 000 seconds\\  
-max_seconds=342000 +\\ 
-restart_mode ='restart'+Initial first job:\\ 
 +max_seconds=162000\\  
 +restart_mode ='from_scratch'\\  
 +\\  
 +Subsequent jobs (2nd, 3rd etc):\\ 
 +max_seconds=162000\\  
 +restart_mode ='restart'\\  
 +\\  
 +\\ 
 +**3.** Your job may not complete in the maximum PBS walltime of 96 hours on the //smp// queue, so then choose choose max_seconds equivalent to less than 96 hours, such as 93 hours:
  
 +93 hours = 93 x 60 x 60 = 334 800 seconds\\ 
 +max_seconds=334800\\ 
 +restart_mode ='from_scratch'\\ 
 +\\ 
 +Subsequent jobs (2nd, 3rd etc):\\
 +max_seconds=334800\\ 
 +restart_mode ='restart'\\ 
  
 +//__**Please remember to delete any unwanted QE restart files and unwanted temporary QE files! We need to do our best with lustre/lustre3p.**__//
/app/dokuwiki/data/attic/howto/quantum_espresso_checkpointing_restarting.1633429587.txt.gz · Last modified: 2021/12/09 16:42 (external edit)

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Noncommercial-Share Alike 3.0 Unported
CC Attribution-Noncommercial-Share Alike 3.0 Unported Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki