Are there any reasons to reduce the simulation duration while re-simulating the same Mfix file with the same geometry, boundary conditions, and time step? if yes then what is the reason behind this?
Because I repeat the simulation, it takes less time to repeat it first.
How can I check the simulation conditions like CPU, GPU, processor use, and core?
You could see some small variation in the time it takes to complete the simulation but it should be minimal (a few percents) if the setup and hardware is exactly the same. If you are using a HPC cluster, please consult with your IT. They can check the load of the nodes to see if there are rogue processes that take up some of the CPU time.
Dear Jeff,
I am not using cluster based simulation, and want to know the reason why simulation time reducing. I simulated one case and time different coming 30 min.
I also want to the know, How can I check the simulation conditions like CPU, GPU, processor use, and core?
Please provide more details: Your operating system, MFiX version, what model (TFM, DEM etc). Are you running in serial, DMP,SMP mode? How many cores are you asking for when you run and how many do you have available on your machine?
I am eager to know my first question reason to reduce simulation time.
And second answer is: i am using 16 core cpu, windows 11, Mfix-23. 4, TFM used, using custom solver, don’t know SMP mode, running serial don’t know( how to check), and i want to know how many cores a simulation are using?
Sorry I cannot answer this question. If you built the custom solver, you should know if you built the SMP solver or the serial solver (look at the build options you selected). If you built the serial solver, then you are using one core. If you built the SMP solver, you should know many cores you are asking for when you launch the simulation (look at the run panel). Please attach you simulation file if you need someone to provide further assistance.
Dear Jeff,
As I checked, I build a custom solver in parallel.
I don’t understand how much time the actual simulation is taking, because I checked the CPU time used shows 6.92 hours, while I saw that it took me 14 hours to complete the task. Can you please clarify these things?
I attached the picture herewith.
.
also, find the simulation file.
lungu44.mfx (10.9 KB)
There is an issue with the reported wall time. We are looking into it. Please do not rely on the displayed wall time, but look at the time stamp of say the vtk files if you want to know the actual simulation time (look at the file properties in your Windows file explorer).
Now I think you may be confused with the build options. If you have “Build solver in parallel” checked, it does not mean you are building a solver that will run on multiple cores (SMP or DMP). It means the Fortran files will be compiled using several cores. This only makes the compilation (build process) faster than if each file was compiled with a single core. If “Distributed memory parallel (DMP)” and “Shared memory parallel (SMP)” are not checked, you are still building a serial solver that will run on a single core. Please note that building the DMP solver is not available on Windows OS.
