Cannot use mpco recorder in OpenSeesSP when running with multiple cores

[STKO Team]

Besides, I am also concerned with performance. For large models (fiber+shell+solid elements), do you think it will have a minor performance improvement using OpenSeesMP which excludes the sendSelf/recvSelf operations?

There is a minor performance improvement in using OpenSeesMP, but it is not due to the fact that it does not use the sendSelf/recvSelf operations, because the partitioning happens only once.

The main point of not using the sendSelf/recvSelf operations is that you can use any element/section/materials, even those who do not implement those operations.

The minor performance improvement is due to the fact that with OpenSeesSP is that the model is already partitioned, and each process will only build its own part of the model.
On the contrary, OpenSeesSP builds the full model into the process P0. Then, when you call "analyze" for the first time, it will do the partition, and send each model part to its process.
If your model is so large that it does not fit 1 process (for RAM memory issue for example) you won't be able to run it.

[jfhuang]

I see, thank you very much for this very helpful explanation!!

[STKO Team]

You're welcome! Good luck with your model.