[gmx-developers] Simulations on GPU

Szilárd Páll szilard.pall at cbr.su.se
Tue Oct 5 10:15:18 CEST 2010


I think this mail belongs to the user's list, CC-d will continue the
discussion there.

--
Szilárd



2010/10/5 Igor Leontyev <ileontyev at ucdavis.edu>:
> Dear gmx-developers,
> My first attempt to start GPU-version of gromacs has no success. The reason
> is that grompp turns off setting of electrostatics overriding them by
> coulombtype=Cut-off and rcoulomb=1, which are not currently supported in gpu
> version, despite that in mdp-file coulombtype is set to "PME" and
> rcoulomb=1.2. See the log-file bellow. Any suggestions?
>
> 1) Here a precompiled mdrun-gpu binary from the distributed archive
> "gromacs-4.5-GPU-beta2_linux-X86_64.tar.gz" was used along with standard
> installation of gromacs version 4.5-beta2. I tried to follow the
> instructions given at
> http://www.gromacs.org/Downloads/Installation_Instructions/Gromacs_on_GPUs
> The output always points to unsupported method of electrostatics
> disregarding the fact that both tried methods (PME and NoCutoff) should be
> supported.
>
> 2) I also tried upper versions of grompp: 4.5-beta3, 4.5-beta4, 4.5 and
> 4.5.1.
> In 4.5-beta3 version grompp does not switch the electrostatics method but
> mdrun-gpu yet generate an error:
>>
>> The requested platform "CUDA" could not be found
>
> While all upper gmx-versions are not compatible with this mdrun-gpu
> executable
>>
>> Fatal error:
>> reading tpx file (CcO_0hcav_3Dch_E242side_OH-_O2-_PM.tpr) version 73 with
>> version 71
>
> Thanks,
> Igor
>
>
>
> The output of gmx4.5-beta2 is
> --------------------------------------------------------------------------
> Log file opened on Mon Oct  4 21:56:10 2010
> Host: powerpc  pid: 30370  nodeid: 0  nnodes:  1
> The Gromacs distribution was built @TMP_TIME@ by
> rossen at mohawk.cbr.su.se [CMAKE] (@TMP_MACHINE@)
>
>                        :-)  G  R  O  M  A  C  S  (-:
>                     GROwing Monsters And Cloning Shrimps
>                       :-)  VERSION 4.5-GPU-beta2  (-:
>
>     Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
>      Copyright (c) 1991-2000, University of Groningen, The Netherlands.
>            Copyright (c) 2001-2008, The GROMACS development team,
>           check out http://www.gromacs.org for more information.
>        This program is free software; you can redistribute it and/or
>         modify it under the terms of the GNU General Public License
>        as published by the Free Software Foundation; either version 2
>            of the License, or (at your option) any later version.
> :-)  /home/leontyev/programs/bin/gromacs/gromacs-4.5-beta2/bin/mdrun-gpu (-:
>
> ++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++
> B. Hess and C. Kutzner and D. van der Spoel and E. Lindahl
> GROMACS 4: Algorithms for highly efficient, load-balanced, and scalable
> molecular simulation
> J. Chem. Theory Comput. 4 (2008) pp. 435-447
> -------- -------- --- Thank You --- -------- --------
>
> ++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++
> D. van der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E. Mark and H. J. C.
> Berendsen
> GROMACS: Fast, Flexible and Free
> J. Comp. Chem. 26 (2005) pp. 1701-1719
> -------- -------- --- Thank You --- -------- --------
>
> ++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++
> E. Lindahl and B. Hess and D. van der Spoel
> GROMACS 3.0: A package for molecular simulation and trajectory analysis
> J. Mol. Mod. 7 (2001) pp. 306-317
> -------- -------- --- Thank You --- -------- --------
>
> ++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++
> H. J. C. Berendsen, D. van der Spoel and R. van Drunen
> GROMACS: A message-passing parallel molecular dynamics implementation
> Comp. Phys. Comm. 91 (1995) pp. 43-56
> -------- -------- --- Thank You --- -------- --------
> Input Parameters:
>  integrator           = md
>  nsteps               = 100
>  init_step            = 0
>  nstcalcenergy        = 10
>  ns_type              = Simple
>  nstlist              = 0
>  ndelta               = 2
>  nstcomm              = 1003
>  comm_mode            = Linear
>  nstlog               = 10
>  nstxout              = 5000
>  nstvout              = 10000000
>  nstfout              = 0
>  nstenergy            = 1000
>  nstxtcout            = 1
>  init_t               = 0
>  delta_t              = 0.001
>  xtcprec              = 1000
>  nkx                  = 0
>  nky                  = 0
>  nkz                  = 0
>  pme_order            = 4
>  ewald_rtol           = 1e-05
>  ewald_geometry       = 0
>  epsilon_surface      = 0
>  optimize_fft         = FALSE
>  ePBC                 = xyz
>  bPeriodicMols        = FALSE
>  bContinuation        = FALSE
>  bShakeSOR            = FALSE
>  etc                  = No
>  nsttcouple           = -1
>  epc                  = No
>  epctype              = Isotropic
>  nstpcouple           = -1
>  tau_p                = 1
>  ref_p (3x3):
>     ref_p[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>     ref_p[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>     ref_p[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>  compress (3x3):
>     compress[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>     compress[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>     compress[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>  refcoord_scaling     = No
>  posres_com (3):
>     posres_com[0]= 0.00000e+00
>     posres_com[1]= 0.00000e+00
>     posres_com[2]= 0.00000e+00
>  posres_comB (3):
>     posres_comB[0]= 0.00000e+00
>     posres_comB[1]= 0.00000e+00
>     posres_comB[2]= 0.00000e+00
>  andersen_seed        = 815131
>  rlist                = 1
>  rlistlong            = 1
>  rtpi                 = 0.05
>  coulombtype          = Cut-off
>  rcoulomb_switch      = 0
>  rcoulomb             = 1
>  vdwtype              = Cut-off
>  rvdw_switch          = 0
>  rvdw                 = 1
>  epsilon_r            = 1
>  epsilon_rf           = 1
>  tabext               = 1
>  implicit_solvent     = No
>  gb_algorithm         = Still
>  gb_epsilon_solvent   = 80
>  nstgbradii           = 1
>  rgbradii             = 1
>  gb_saltconc          = 0
>  gb_obc_alpha         = 1
>  gb_obc_beta          = 0.8
>  gb_obc_gamma         = 4.85
>  gb_dielectric_offset = 0.009
>  sa_algorithm         = No
>  sa_surface_tension   = 2.092
>  DispCorr             = No
>  free_energy          = no
>  init_lambda          = 0
>  delta_lambda         = 0
>  n_foreign_lambda     = 0
>  sc_alpha             = 0
>  sc_power             = 0
>  sc_sigma             = 0.3
>  nstdhdl              = 10
>  dh_table_size        = 0
>  dh_table_spacing     = 0.1
>  nwall                = 0
>  wall_type            = 9-3
>  wall_atomtype[0]     = -1
>  wall_atomtype[1]     = -1
>  wall_density[0]      = 0
>  wall_density[1]      = 0
>  wall_ewald_zfac      = 3
>  pull                 = no
>  disre                = No
>  disre_weighting      = Conservative
>  disre_mixed          = FALSE
>  dr_fc                = 1000
>  dr_tau               = 0
>  nstdisreout          = 100
>  orires_fc            = 0
>  orires_tau           = 0
>  nstorireout          = 100
>  dihre-fc             = 1000
>  em_stepsize          = 0.01
>  em_tol               = 10
>  niter                = 20
>  fc_stepsize          = 0
>  nstcgsteep           = 1000
>  nbfgscorr            = 10
>  ConstAlg             = Lincs
>  shake_tol            = 0.0001
>  lincs_order          = 8
>  lincs_warnangle      = 30
>  lincs_iter           = 4
>  bd_fric              = 0
>  ld_seed              = 1993
>  cos_accel            = 0
>  deform (3x3):
>     deform[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>     deform[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>     deform[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
>  userint1             = 0
>  userint2             = 0
>  userint3             = 0
>  userint4             = 0
>  userreal1            = 0
>  userreal2            = 0
>  userreal3            = 0
>  userreal4            = 0
> grpopts:
>  nrdf:       99024
>  ref_t:           0
>  tau_t:           0
> anneal:          No
> ann_npoints:           0
>  acc:            0           0           0
>  nfreeze:           N           N           N
>  energygrp_flags[  0]: 0 0
>  energygrp_flags[  1]: 0 0
>  efield-x:
>     n = 0
>  efield-xt:
>     n = 0
>  efield-y:
>     n = 0
>  efield-yt:
>     n = 0
>  efield-z:
>     n = 0
>  efield-zt:
>     n = 0
>  bQMMM                = FALSE
>  QMconstraints        = 0
>  QMMMscheme           = 0
>  scalefactor          = 1
> qm_opts:
>  ngQM                 = 0
> Table routines are used for coulomb: FALSE
> Table routines are used for vdw:     FALSE
> Cut-off's:   NS: 1   Coulomb: 1   LJ: 1
> System total charge: -0.000
> Generated table with 1000 data points for 1-4 COUL.
> Tabscale = 500 points/nm
> Generated table with 1000 data points for 1-4 LJ6.
> Tabscale = 500 points/nm
> Generated table with 1000 data points for 1-4 LJ12.
> Tabscale = 500 points/nm
> Enabling SPC-like water optimization for 11505 molecules.
> Configuring nonbonded kernels...
> Configuring standard C nonbonded kernels...
>
> Removing pbc first time
> Initializing LINear Constraint Solver
> ++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++
> B. Hess and H. Bekker and H. J. C. Berendsen and J. G. E. M. Fraaije
> LINCS: A Linear Constraint Solver for molecular simulations
> J. Comp. Chem. 18 (1997) pp. 1463-1472
> -------- -------- --- Thank You --- -------- --------
> The number of constraints is 5967
> ++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++
> S. Miyamoto and P. A. Kollman
> SETTLE: An Analytical Version of the SHAKE and RATTLE Algorithms for Rigid
> Water Models
> J. Comp. Chem. 13 (1992) pp. 952-962
> -------- -------- --- Thank You --- -------- --------
> Center of mass motion removal mode is Linear
> We have the following groups for center of mass motion removal:
>  0:  rest
> Max number of connections per atom is 91
> Total number of connections is 387700
> Max number of graph edges per atom is 6
> Total number of graph edges is 70330
> OpenMM plugins loaded from directory
> /home/leontyev/programs/bin/gromacs/gromacs-4.5-beta2/bin/../openmm/lib/plugins:
> libOpenMMCuda.so,
> -------------------------------------------------------
> Program mdrun-gpu, VERSION 4.5-GPU-beta2
> Source code file:
> /home/rossen/Research/Projects/Gromacs-dev/gromacs/src/kernel/openmm_wrapper.cpp,
> line: 550
> Fatal error:
> OpenMM supports only the following methods for electrostatics: NoCutoff
> (i.e. rcoulomb = rvdw = 0 ),Reaction-Field, Ewald or PME.
> For more information and tips for troubleshooting, please check the GROMACS
> website at http://www.gromacs.org/Documentation/Errors
> -------------------------------------------------------
> --
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the www interface
> or send it to gmx-developers-request at gromacs.org.
>



More information about the gromacs.org_gmx-developers mailing list