Distributed-memory parallel processing

- November 19, 2024 at 9:06 am
  
  2020221038
  Subscriber
  
  When distributed-memory parallel processing is used, the merging of individual small result files takes too long because the result files are too large. Closing it directly through the task manager and viewing the results again results in an error. How can this problem be solved?
- November 19, 2024 at 12:37 pm
  
  peteroznewman
  Subscriber
  
  The default setting for merging multiple .rst files from the distributed solve is to not do that. Workbench can display results without doing that. Have you tried not merging the .rst files?
  - November 19, 2024 at 12:51 pm
    
    2020221038
    Subscriber
    
    Every time it shows that the solution is complete, it automatically starts merging. fangan-1118-2.rst is the result file that is merged, and if it's merged without completing the merge, it's size is 0KB. the results of importing fangan-1118-2_0.rst (unmerged file) into my model show the result as an error. How can I set up not to merge the result or show the result in workbench? I am mainly using APDL.
    - November 19, 2024 at 10:22 pm
      
      peteroznewman
      Subscriber
      
      In Mechanical, here is where you tell the solver what to do with the restart files:
      Consult the APDL manual for the solver parameter to stop that. Maybe someone who knows will reply.
    - May 9, 2025 at 2:46 am
      
      2020221038
      Subscriber
      
      Okay, thanks! I've benefited from your answers.