I’m attempting to do a millisecond to second range Coarse Grained MD with gromacs. I have both Martini22p and Martini3.0.0 systems of protein-membrane-solvent already minimised and equilibrated. When I run the MD with mdrun, I get a lot of problems with load imbalance and millisecond range simulations would take months.
Does anyone use any other method for Coarse Grained simulations in gromacs? Where do you build the system?
How are you building the membrane-protein system (insane, CHARMM-GUI, any other?)? What are the error messages? How big is the system?
Since you mentioned it is a membrane-protein system (and I assume there will be water and ions, too), this will be atleast 100K atom/bead system. Running 1 ms simulation can take weeks depending on your system configuration. If you want speedup, you may consider shifting to Martini 2.0/2.2.
Thank you for your reply! I used CHARMM-GUI Martini Maker and relaying on my 1 microsecond simulation cost, I should wait for about 60 days for my 1 millisecond simulation. But this comes along with your recommendations as I could actually try running my jobs in a better computer and also optimise the partitioning of the system. I was just wondering if there was any other method I was not considering as I have no reference to compare to.
Also, does anyone know any command to improve domain decomposition and load imbalance apart from -dd -rdd or -ntomp to manually adjust the partitioning? I am using reaction-field electrostatics instead of PME.