|
[Sponsors] |
June 7, 2018, 14:20 |
Data stored after first fvm::laplacian/div
|
#1 |
New Member
Ville
Join Date: Jun 2018
Posts: 2
Rep Power: 0 |
Hello,
Currently, I am investigating the possibility of optimizing a code. More prescisely, looking at the way OpenFOAM handles differential operators. I took a tutorial case and clocked the time it took the solver to run over some differential operators at each timestep. I clocked fvm::laplacian(U) at the start of each timestep with the velocity field getting updated between timesteps. First, the fvm::laplacian(U) runs in roughly 110 tics, after which it drops down to a fifth (~20 tics) for the subsequent timesteps. After this, I added a fvm::div(phi,U) in the very beginning of the solver, right after the initialization of fields. By running the code again the tics were ~45 (divergence), ~85 (first laplacian) and ~20 (subsequent laplacians). Something is clearly stored after the first differential operators which accounts for a shorter run time (data regarding the matrix?) However, by tracing the files, particularly fvmLaplacian, gaussLaplacianScheme and fmvDiv. However, I cannot understand what is going on... So, essentially my question is what data is stored for future use when these are used the first time? Br. Ville |
|
Tags |
fvm::div, fvm::laplacian, stored data |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
paraview - cell data or point data on plot over line | bye bye my blue | OpenFOAM | 0 | December 13, 2016 07:07 |
【Help】"Error: Update_Time_Level: invalid data" | Chen | FLUENT | 2 | August 24, 2014 08:51 |
[General] Warping Cells by Vectors? | ronwelty | ParaView | 10 | April 29, 2014 06:52 |
Problem in running ICEM grid in Openfoam | Tarak | OpenFOAM | 6 | September 9, 2011 18:51 |
CGNS vs Tecplot Data Format | LWhitson2 | Main CFD Forum | 3 | July 1, 2011 14:50 |