Hello,
I was trying to run pvbatch in parallel but it seems that it only works on one rank. I tried both with the shipped mpiexec in the /bin directory and the one already present on my desktop but the behaviour is the same.
What am I missing?
Is your data distributed ?
What are you referring to in particular?
I was trying to run pvbatch in parallel but it seems that it only works on one rank.
Please provide more info about what kind of work you are doing and what kind of data you are opening.
pvserver is working distributed if and only if the data itself is distributed.
Please provide more info about what kind of work you are doing and what kind of data you are opening.
It is executing a paraview trace with pvbatch on a .vtu script (Paraview 5.12)
A .vtu file is not distributed natively. You can see that by displaying the “vtkProcessId” field.
Add a “DistributeDataSet” filter to your script to distribute the data.
Thanks a lot for the insight!
Add a “DistributeDataSet” filter to your script to distribute the data.
I only have Distribute Points, is this?
Sorry, the name is “Redistribute DataSet”
Ok now it works! Thanks a lot!
Any filter option do you recommend using for better performance?
Best
Redistribute your data only once and save it to a distributed format like .pvtu, it will then be much more performant as I/O will be distributed too.
But TBH this is a very complex topic: https://www.kitware.com/paraview-hpc-scalability-guide/