The more mpi processes the slower the app Walberla/walberla-v6.1/apps/tutorials/lbm/03_LBLidDrivenCavity.cpp
Hi, i have been trying to run the Walberla example for the LidDrivenCavity example, as located in Walberla/walberla-v6.1/apps/tutorials/lbm/03_LBLidDrivenCavity.cpp. After compiling, to run i used, mpirun -np 1 ./03_LBLidDrivenCavity the runtime was 4.402 sec. When i increased the number of processes to 4 then the runtime i got was 29.139 sec, increasing processes to 16 the runtime i got was 41.262 sec. Under normal circumstances as i increase processes, i.e time would decrease. How can performance improve as I increase the processes? Am i missing something in the mpirun call?
Edited by mosemb