# Running Gromacs in GPU vs CPU **URL:** \ **Category:** User discussions\ **Tags:** mdp-parameters, mdrun, gpu\ **Created:** [December 1, 2021, 7:56am UTC](https://gromacs.bioexcel.eu/t/running-gromacs-in-gpu-vs-cpu/3230 "2021-12-01T07:56:10Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![Shantanu](https://avatars.discourse-cdn.com/v4/letter/s/b19c9b/32.png) [@Shantanu](https://gromacs.bioexcel.eu/u/Shantanu)\ **Post date:** [December 1, 2021, 7:56am UTC](https://gromacs.bioexcel.eu/t/running-gromacs-in-gpu-vs-cpu/3230/1 "2021-12-01T07:56:10Z")
GROMACS version:2020.6 GROMACS modification: No 1.Why changes in rlist happen when I run the same thing in a GPU vs CPU? 2.I have set the r-list and nst-list value in the parameters files , yet why they change. And do these changes has any impact on the results? 1. GPU gmx mdrun -deffnm complex\_nvt -v -nt 64 Reading file complex\_nvt.tpr, VERSION 2020.2-dev-20200430-5e78835-unknown (single precision) Changing nstlist from 5 to 100, rlist from 1.2 to 1.346 1 GPU selected for this run. Mapping of GPU IDs to the 2 GPU tasks in the 1 rank on this node: PP:0,PME:0 PP tasks will do (non-perturbed) short-ranged interactions on the GPU PP task will update and constrain coordinates on the CPU PME tasks will do all aspects on the GPU Using 1 MPI thread Non-default thread affinity set, disabling internal thread affinity Using 64 OpenMP threads 1. CPU gmx mdrun -deffnm complex\_nvt -v Reading file complex\_nvt.tpr, VERSION 2020.6 (single precision) Changing nstlist from 5 to 80, rlist from 1.2 to 1.322 Using 1 MPI thread Non-default thread affinity set, disabling internal thread affinity Using 12 OpenMP threads These are my parameters- nvt- vi nvt title = NVT define = -DPOSRES ; position restrain the protein ; Run parameters integrator = md ; leap-frog integrator nsteps = 500000 ; 2 \* 500000 = 1 ns dt = 0.002 ; 2 fs ; Output control nstxout = 100 ; save coordinates every 0.2 ps nstvout = 100 ; save velocities every 0.2 ps nstenergy = 100 ; save energies every 0.2 ps nstlog = 100 ; update log file every 0.2 ps ; Bond parameters continuation = no ; first dynamics run constraint\_algorithm = lincs ; holonomic constraints constraints = h-bonds ; bonds involving H are constrained lincs\_iter = 1 ; accuracy of LINCS lincs\_order = 4 ; also related to accuracy ; Neighborsearching ns\_type = grid ; search neighboring grid cels nstlist = 5 ; 10 fs rlist = 1.2 ; short-range neighborlist cutoff (in nm) rcoulomb = 1.2 ; short-range electrostatic cutoff (in nm) rvdw = 1.2 ; short-range van der Waals cutoff (in nm) cutoff-scheme = Verlet ; Electrostatics coulombtype = PME ; Particle Mesh Ewald for long-range electrostatics pme\_order = 4 ; cubic interpolation fourierspacing = 0.16 ; grid spacing for FFT ; Temperature coupling is on tcoupl = V-rescale ; modified Berendsen thermostat tc-grps = ATG4B LC3B SOL\_NA ; two coupling groups - more accurate tau\_t = 0.1 0.1 0.1 ; time constant, in ps ref\_t = 310 310 310 ; reference temperature, one for each group, in K ; Pressure coupling is off pcoupl = no ; no pressure coupling in NVT ; Periodic boundary conditions pbc = xyz ; 3-D PBC ; Dispersion correction DispCorr = EnerPres ; account for cut-off vdW scheme ; Velocity generation gen\_vel = yes ; assign velocities from Maxwell distribution gen\_temp = 310 ; temperature for Maxwell distribution gen\_seed = -1 ; generate a random seed ; COM motion removal ; These options remove motion of the protein/bilayer relative to the solvent/ions nstcomm = 1 comm-mode = Linear comm-grps = ATG4B LC3B SOL\_NA I am new to gromacs and would really appreciate the help. Thank You in advance! ---
**Author:** ![Masrul](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/masrul/32/234_2.png) [@Masrul](https://gromacs.bioexcel.eu/u/Masrul)\ **Post date:** [December 3, 2021, 6:06am UTC](https://gromacs.bioexcel.eu/t/running-gromacs-in-gpu-vs-cpu/3230/2 "2021-12-03T06:06:40Z")
nst-list determines after how many steps neighborlist gets updated, r-list is how far neighbors are looked for. So, for maintaining accuracy, when one is decreased, other has to be increased. Neighborlist update is expensive due to cpu-gpu communication, that’s why gromacs adjust nst-list/r-list for optimal performance.