Hi, I am trying to reproduce the results you have published here: https://palabos.unige.ch/community/cpu-gpu-80-days/. Have these been published anywhere else with more details? If you are using standard C++ to parallelise, why did you compare new Palabos-Hybrid to the original CPU-parallel version of Palabos and not just re-compile with -stdpar=multicore? I have been having trouble with compiling the tgv3d example for only CPUs myself!