faster way to use apdl, pymapdl

[francesco123212]

Hi everyone,
hope yall are ok!
I have a question related to the speed of the pymapdl package:
which is the fastest way to perform a simulation?

• use only old school apdl
• try to minimize the number of interactions between python/julia and mapdl using lots of run_multiline, e.g., to create a line

pymapdl.run_multiline("""
/prep7
k,1,coordinates
k,2,coordintaes
l,1,2
""")

• only use py mapdl WO run_multiline, e.g., to create a line:

pymapdl.prep7()
pymapdl.k() #create first keypoint
pymapdl.k() # create second kp
pymapdl.l() # create line connecting the two kp

[germa89]

This should be a discussion. Converting...

[germa89]

Short answer

Solving is not affected by PyMAPDL overhead. However sending many PyMAPDL commands is slower than MAPDL commands.

Long answer

Each PyMAPDL command gets converted to an MAPDL command and then sent to the MAPDL instance. This conversion, sending this command, processing it and getting back takes time. If we compare PyMAPDL and MAPDL commands, the time difference will be only the conversion, which in most of the cases is negligible.

Longest answer

However because PyMAPDL interact with an MAPDL instance (exchanging commands), the more command you send at once the better (you not need to wait for the current command before sending the other one). Hence I can recommend you to use mapdl.non_interactive mode or mapdl.input as much as you can because you will be executing a strem of MAPDL commands in the MAPDL instance without Python overhead or response delays.

However having said that, unless you do a lot of pymapdl commands (like thousands) the overall time will not change much compared a full MAPDL commands. And as I mentioned at the beginning, the most of the time will go to the solving (SOLVE). PyMAPDL impact on the solving time is negligible or zero.
Furthermore, PyMAPDL can save you time in the post-processing stage because of the many function helpers included.

Further reading

I recommend you to have a look at this conversation: #757