Re: [gmx-users] handling particle decomposition with distance restraints
The energy minimization went on without any problem on 4 processors but the problem occurs when I perform the MD run. Also, I did not get any error message with relevance to LINCS etc... JJ On Wed, Jun 24, 2009 at 6:53 PM, Justin A. Lemkul jalem...@vt.edu wrote: jayant james wrote: Yes my distance restraints are long because I am using FRET distances as distance restraints while performing MD simulations. Upon usage of this command mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro -pd I get the following error!! I did try giving yes after -pd but even then the same error message is repeated. Anything else printed to the screen or log file? LINCS warnings or anything else? Did energy minimization complete successfully? -Justin Back Off! I just backed up md.log to ./#md.log.9# Reading file pr.tpr, VERSION 4.0 (single precision) NNODES=4, MYRANK=1, HOSTNAME=localhost.localdomain NODEID=1 argc=13 NODEID=3 argc=13 [localhost:17514] *** Process received signal *** [localhost:17514] Signal: Segmentation fault (11) [localhost:17514] Signal code: Address not mapped (1) [localhost:17514] Failing at address: 0x134 [localhost:17514] [ 0] /lib64/libpthread.so.0 [0x38dec0f0f0] [localhost:17514] [ 1] /lib64/libc.so.6(memcpy+0x15b) [0x38de08432b] [localhost:17514] [ 2] /usr/lib64/openmpi/1.2.4-gcc/libmpi.so.0(ompi_convertor_pack+0x152) [0x3886e45392] [localhost:17514] [ 3] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_btl_sm.so(mca_btl_sm_prepare_src+0x13d) [0x7f39dd118a4d] [localhost:17514] [ 4] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_pml_ob1.so(mca_pml_ob1_send_request_start_rndv+0x140) [0x7f39dd735230] [localhost:17514] [ 5] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_pml_ob1.so(mca_pml_ob1_send+0x748) [0x7f39dd72e508] [localhost:17514] [ 6] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_coll_tuned.so(ompi_coll_tuned_bcast_intra_split_bintree+0x91c) [0x7f39dc6f735c] [localhost:17514] [ 7] /usr/lib64/openmpi/1.2.4-gcc/libmpi.so.0(MPI_Bcast+0x15c) [0x3886e4c40c] [localhost:17514] [ 8] mdrun_mpi(bcast_state+0x26c) [0x56d59c] [localhost:17514] [ 9] mdrun_mpi(mdrunner+0x1067) [0x42b807] [localhost:17514] [10] mdrun_mpi(main+0x3b4) [0x431c34] [localhost:17514] [11] /lib64/libc.so.6(__libc_start_main+0xe6) [0x38de01e576] [localhost:17514] [12] mdrun_mpi [0x413339] [localhost:17514] *** End of error message *** mpirun noticed that job rank 0 with PID 17514 on node localhost.localdomain exited on signal 11 (Segmentation fault). 3 additional processes aborted (not shown) On Wed, Jun 24, 2009 at 6:04 PM, Justin A. Lemkul jalem...@vt.edumailto: jalem...@vt.edu wrote: jayant james wrote: I just replaced the old gmx 4.0 version with the 4.0.5 version and still the same problem NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) Well, does it work with -pd? It looks like your distance restraints are indeed quite long, so this looks like it is your only option. -Justin WARNING: Can not write distance restraint data to energy file with domain decomposition On Wed, Jun 24, 2009 at 5:10 PM, Justin A. Lemkul jalem...@vt.edu mailto:jalem...@vt.edu mailto:jalem...@vt.edu mailto:jalem...@vt.edu wrote: jayant james wrote: Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Upgrade to the latest version (4.0.5), since there have been numerous improvements to domain decomposition throughout the development of version 4.0. -Justin Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance
[gmx-users] handling particle decomposition with distance restraints
Why not use the pull code? If you haev to use distance restraints, then try LAM mpi with your pd run. We had similar error messages with vanilla .mdp files using openmpi with large and complex systems that went away when we switched to LAM MPI. Our problems disappeared in gmx 4 so we went back to openmpi for all systems as that mdrun_mpi version is faster in our hands. I admit, there is no good reason why LAM would work and openMPI would not, but I have seen it happen before so it's worth a shot. -- original message-- The energy minimization went on without any problem on 4 processors but the problem occurs when I perform the MD run. Also, I did not get any error message with relevance to LINCS etc... JJ ___ gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] handling particle decomposition with distance restraints
Let me re-emphasize that the pull code may be a good solution for you. As per your request, I currently use the following without any problems: fftw 3.1.2 gromacs 3.3.1 or 4.0.4 openmpi 1.2.6 Be especially aware that openmpi 1.3.0 and 1.3.1 are broken, as I posted here: http://lists.gromacs.org/pipermail/gmx-users/2009-March/040844.html To be clear, I have never experienced any openmpi-based problems with any version of gromacs 4 and openmpi 1.2.6. I posted the original notice of our problems with openmpi (1.2.1) that were solved by using lam here. http://www.mail-archive.com/gmx-users@gromacs.org/msg08257.html Chris jayant james wrote: Hi! thanks for your mail. Could you please share what OS and versions of fftw, openmpi and gmx you are currently using. Thanks you JJ On Thu, Jun 25, 2009 at 12:28 PM, chris.ne...@utoronto.ca mailto:chris.ne...@utoronto.ca wrote: Why not use the pull code? If you haev to use distance restraints, then try LAM mpi with your pd run. We had similar error messages with vanilla .mdp files using openmpi with large and complex systems that went away when we switched to LAM MPI. Our problems disappeared in gmx 4 so we went back to openmpi for all systems as that mdrun_mpi version is faster in our hands. I admit, there is no good reason why LAM would work and openMPI would not, but I have seen it happen before so it's worth a shot. -- original message-- The energy minimization went on without any problem on 4 processors but the problem occurs when I perform the MD run. Also, I did not get any error message with relevance to LINCS etc... JJ ___ gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use thewww interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php -- Jayasundar Jayant James www.chick.com/reading/tracts/0096/0096_01.asp http://www.chick.com/reading/tracts/0096/0096_01.asp) ___ gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] handling particle decomposition with distance restraints
We are running SUSE Linux It's best to keep all of this on the mailing list in case it becomes useful to somebody else. jayant james wrote: Hi! thanks for your mail. I have never used pull code before and so I am a bit apprehensive but I do accept your suggestion and I am working on that. By the wat what is the OS that you are using? Is it Suse or Fedora? Thanks Jayant James On Thu, Jun 25, 2009 at 4:31 PM, Chris Neale chris.ne...@utoronto.ca mailto:chris.ne...@utoronto.ca wrote: Let me re-emphasize that the pull code may be a good solution for you. As per your request, I currently use the following without any problems: fftw 3.1.2 gromacs 3.3.1 or 4.0.4 openmpi 1.2.6 Be especially aware that openmpi 1.3.0 and 1.3.1 are broken, as I posted here: http://lists.gromacs.org/pipermail/gmx-users/2009-March/040844.html To be clear, I have never experienced any openmpi-based problems with any version of gromacs 4 and openmpi 1.2.6. I posted the original notice of our problems with openmpi (1.2.1) that were solved by using lam here. http://www.mail-archive.com/gmx-users@gromacs.org/msg08257.html Chris jayant james wrote: Hi! thanks for your mail. Could you please share what OS and versions of fftw, openmpi and gmx you are currently using. Thanks you JJ On Thu, Jun 25, 2009 at 12:28 PM, chris.ne...@utoronto.ca mailto:chris.ne...@utoronto.ca mailto:chris.ne...@utoronto.ca mailto:chris.ne...@utoronto.ca wrote: Why not use the pull code? If you haev to use distance restraints, then try LAM mpi with your pd run. We had similar error messages with vanilla .mdp files using openmpi with large and complex systems that went away when we switched to LAM MPI. Our problems disappeared in gmx 4 so we went back to openmpi for all systems as that mdrun_mpi version is faster in our hands. I admit, there is no good reason why LAM would work and openMPI would not, but I have seen it happen before so it's worth a shot. -- original message-- The energy minimization went on without any problem on 4 processors but the problem occurs when I perform the MD run. Also, I did not get any error message with relevance to LINCS etc... JJ ___ gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org mailto:gmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use thewww interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php -- Jayasundar Jayant James www.chick.com/reading/tracts/0096/0096_01.asp http://www.chick.com/reading/tracts/0096/0096_01.asp http://www.chick.com/reading/tracts/0096/0096_01.asp) -- Jayasundar Jayant James www.chick.com/reading/tracts/0096/0096_01.asp http://www.chick.com/reading/tracts/0096/0096_01.asp) ___ gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
[gmx-users] handling particle decomposition with distance restraints
Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) WARNING: Can not write distance restraint data to energy file with domain decomposition --- Program mdrun_mpi, VERSION 4.0.2 Source code file: domdec.c, line: 5842 Fatal error: There is no domain decomposition for 4 nodes that is compatible with the given box and a minimum cell size of 9.85926 nm Change the number of nodes or mdrun option -rdd or -dds Look in the log file for details on the domain decomposition --- *pr.mdp* ; User spoel (236) ; Wed Nov 3 17:12:44 1993 ; Input file ; title = Yo cpp = /usr/bin/cpp define = -DDISRES constraints = none ;constraint_algorithm = lincs ;lincs_order = 4 integrator = md dt = 0.001; ps ! nsteps = 400 ; total 2.0ns. nstcomm = 1 nstxout = 5 nstvout = 5 nstfout = 5 nstlog = 5 nstenergy = 500 nstlist = 10 ns_type = grid rlist = 1.0 coulombtype = PME rcoulomb= 1.0 vdwtype = cut-off rvdw= 1.4 fourierspacing = 0.12 fourier_nx = 0 fourier_ny = 0 fourier_nz = 0 pme_order = 4 ewald_rtol = 1e-5 optimize_fft= yes disre = simple disre_weighting = equal ; Berendsen temperature coupling is on in two groups Tcoupl = V-rescale tc-grps = Protein Non-Protein tau_t = 0.1 0.1 ref_t = 300 300 ; Energy monitoring energygrps = Protein Non-Protein ;tnc Non-Protein tnt NMR tni ; Pressure coupling is not on Pcoupl = parrinello-rahman tau_p = 0.5 compressibility = 4.5e-5 ref_p = 1.0 ;simulated annealing ;Type of annealing form each temperature group (no/single/periodic) ;annealing = no, no, no, single, no ; ;Number of annealing points to use for specifying annealing in each group ;annealing_npoints = 0, 0, 0, 9, 0 ; ; List of times at the annealing points for each group ;annealing_time = 0 25 50 75 100 125 150 175 200 ; Temp.at each annealing point, for each group. ;annealing_temp = 300 350 400 450 500 450 400 350 300 * distance restraints file* distance_restraints ] ; ai aj typeindex type' low up1 up2 fac ;TnT240-TnI131, 145, 151, 160, 167 (ca+-7) 201938891 1 1 3.913.915.31 0.574679 201940561 2 1 4.864.866.26 0.409911 201941331 3 1 5.695.697.09 0.457947 201942071 4 1 6.636.638.03 0.323852 201942731 5 1 7.147.148.54 0.294559 ;TnT276- Tni 131,145,151,160,167,5,17,27,40 243438891 6 1 1.341.342.74 4.884769 243440561 7 1 2.132.133.53 0.523368 243441331 8 1 3.663.665.06 0.409911 243442071 9 1 4.484.485.88 0.547825 243442731 10 1 5.435.436.83 0.285938 243426281 11 1 5.895.897.29 0.241333 243427191 12 1 4.764.766.16 0.366358 243428241 13 1 3.813.815.21 0.644145 243429721 14 1 3.103.104.50 0.431009 ;TnT288- Tni 131,145,151,160,167,5,17,27,40 25573889
Re: [gmx-users] handling particle decomposition with distance restraints
jayant james wrote: Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Upgrade to the latest version (4.0.5), since there have been numerous improvements to domain decomposition throughout the development of version 4.0. -Justin Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) WARNING: Can not write distance restraint data to energy file with domain decomposition --- Program mdrun_mpi, VERSION 4.0.2 Source code file: domdec.c, line: 5842 Fatal error: There is no domain decomposition for 4 nodes that is compatible with the given box and a minimum cell size of 9.85926 nm Change the number of nodes or mdrun option -rdd or -dds Look in the log file for details on the domain decomposition --- *pr.mdp* ; User spoel (236) ; Wed Nov 3 17:12:44 1993 ; Input file ; title = Yo cpp = /usr/bin/cpp define = -DDISRES constraints = none ;constraint_algorithm = lincs ;lincs_order = 4 integrator = md dt = 0.001; ps ! nsteps = 400 ; total 2.0ns. nstcomm = 1 nstxout = 5 nstvout = 5 nstfout = 5 nstlog = 5 nstenergy = 500 nstlist = 10 ns_type = grid rlist = 1.0 coulombtype = PME rcoulomb= 1.0 vdwtype = cut-off rvdw= 1.4 fourierspacing = 0.12 fourier_nx = 0 fourier_ny = 0 fourier_nz = 0 pme_order = 4 ewald_rtol = 1e-5 optimize_fft= yes disre = simple disre_weighting = equal ; Berendsen temperature coupling is on in two groups Tcoupl = V-rescale tc-grps = Protein Non-Protein tau_t = 0.1 0.1 ref_t = 300 300 ; Energy monitoring energygrps = Protein Non-Protein ;tnc Non-Protein tnt NMR tni ; Pressure coupling is not on Pcoupl = parrinello-rahman tau_p = 0.5 compressibility = 4.5e-5 ref_p = 1.0 ;simulated annealing ;Type of annealing form each temperature group (no/single/periodic) ;annealing = no, no, no, single, no ; ;Number of annealing points to use for specifying annealing in each group ;annealing_npoints = 0, 0, 0, 9, 0 ; ; List of times at the annealing points for each group ;annealing_time = 0 25 50 75 100 125 150 175 200 ; Temp.at each annealing point, for each group. ;annealing_temp = 300 350 400 450 500 450 400 350 300 * distance restraints file* distance_restraints ] ; ai aj typeindex type' low up1 up2 fac ;TnT240-TnI131, 145, 151, 160, 167 (ca+-7) 201938891 1 1 3.913.915.31 0.574679 201940561 2 1 4.864.866.26 0.409911 201941331 3 1 5.695.697.09 0.457947 201942071 4 1 6.636.638.03 0.323852 201942731 5 1 7.147.148.54 0.294559 ;TnT276- Tni 131,145,151,160,167,5,17,27,40 243438891 6 1 1.341.342.74 4.884769 243440561 7 1 2.132.133.53 0.523368 243441331 8 1 3.663.665.06 0.409911 24344207
Re: [gmx-users] handling particle decomposition with distance restraints
I just replaced the old gmx 4.0 version with the 4.0.5 version and still the same problem NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) WARNING: Can not write distance restraint data to energy file with domain decomposition On Wed, Jun 24, 2009 at 5:10 PM, Justin A. Lemkul jalem...@vt.edu wrote: jayant james wrote: Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Upgrade to the latest version (4.0.5), since there have been numerous improvements to domain decomposition throughout the development of version 4.0. -Justin Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) WARNING: Can not write distance restraint data to energy file with domain decomposition --- Program mdrun_mpi, VERSION 4.0.2 Source code file: domdec.c, line: 5842 Fatal error: There is no domain decomposition for 4 nodes that is compatible with the given box and a minimum cell size of 9.85926 nm Change the number of nodes or mdrun option -rdd or -dds Look in the log file for details on the domain decomposition --- *pr.mdp* ; User spoel (236) ; Wed Nov 3 17:12:44 1993 ; Input file ; title = Yo cpp = /usr/bin/cpp define = -DDISRES constraints = none ;constraint_algorithm = lincs ;lincs_order = 4 integrator = md dt = 0.001; ps ! nsteps = 400 ; total 2.0ns. nstcomm = 1 nstxout = 5 nstvout = 5 nstfout = 5 nstlog = 5 nstenergy = 500 nstlist = 10 ns_type = grid rlist = 1.0 coulombtype = PME rcoulomb= 1.0 vdwtype = cut-off rvdw= 1.4 fourierspacing = 0.12 fourier_nx = 0 fourier_ny = 0 fourier_nz = 0 pme_order = 4 ewald_rtol = 1e-5 optimize_fft= yes disre = simple disre_weighting = equal ; Berendsen temperature coupling is on in two groups Tcoupl = V-rescale tc-grps = Protein Non-Protein tau_t = 0.1 0.1 ref_t = 300 300 ; Energy monitoring energygrps = Protein Non-Protein ;tnc Non-Protein tnt NMR tni ; Pressure coupling is not on Pcoupl = parrinello-rahman tau_p = 0.5 compressibility = 4.5e-5 ref_p = 1.0 ;simulated annealing ;Type of annealing form each temperature group (no/single/periodic) ;annealing = no, no, no, single, no ; ;Number of annealing points to use for specifying annealing in each group ;annealing_npoints = 0, 0, 0, 9, 0 ; ; List of times at the annealing points for each group ;annealing_time = 0 25 50 75 100 125 150 175 200 ; Temp.at each annealing point, for each group. ;annealing_temp = 300 350 400 450 500 450 400 350 300 * distance restraints file* distance_restraints ] ; ai aj typeindex type' low up1 up2 fac ;TnT240-TnI131, 145, 151, 160, 167 (ca+-7) 201938891 1 1 3.913.915.31 0.574679 201940561 2 1 4.864.866.26 0.409911 201941331 3 1 5.695.697.09 0.457947 201942071 4 1 6.636.638.03 0.323852 201942731 5 1 7.147.14
Re: [gmx-users] handling particle decomposition with distance restraints
jayant james wrote: I just replaced the old gmx 4.0 version with the 4.0.5 version and still the same problem NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) Well, does it work with -pd? It looks like your distance restraints are indeed quite long, so this looks like it is your only option. -Justin WARNING: Can not write distance restraint data to energy file with domain decomposition On Wed, Jun 24, 2009 at 5:10 PM, Justin A. Lemkul jalem...@vt.edu mailto:jalem...@vt.edu wrote: jayant james wrote: Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Upgrade to the latest version (4.0.5), since there have been numerous improvements to domain decomposition throughout the development of version 4.0. -Justin Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) WARNING: Can not write distance restraint data to energy file with domain decomposition --- Program mdrun_mpi, VERSION 4.0.2 Source code file: domdec.c, line: 5842 Fatal error: There is no domain decomposition for 4 nodes that is compatible with the given box and a minimum cell size of 9.85926 nm Change the number of nodes or mdrun option -rdd or -dds Look in the log file for details on the domain decomposition --- *pr.mdp* ; User spoel (236) ; Wed Nov 3 17:12:44 1993 ; Input file ; title = Yo cpp = /usr/bin/cpp define = -DDISRES constraints = none ;constraint_algorithm = lincs ;lincs_order = 4 integrator = md dt = 0.001; ps ! nsteps = 400 ; total 2.0ns. nstcomm = 1 nstxout = 5 nstvout = 5 nstfout = 5 nstlog = 5 nstenergy = 500 nstlist = 10 ns_type = grid rlist = 1.0 coulombtype = PME rcoulomb= 1.0 vdwtype = cut-off rvdw= 1.4 fourierspacing = 0.12 fourier_nx = 0 fourier_ny = 0 fourier_nz = 0 pme_order = 4 ewald_rtol = 1e-5 optimize_fft= yes disre = simple disre_weighting = equal ; Berendsen temperature coupling is on in two groups Tcoupl = V-rescale tc-grps = Protein Non-Protein tau_t = 0.1 0.1 ref_t = 300 300 ; Energy monitoring energygrps = Protein Non-Protein ;tnc Non-Protein tnt NMR tni ; Pressure coupling is not on Pcoupl = parrinello-rahman tau_p = 0.5 compressibility = 4.5e-5 ref_p = 1.0
Re: [gmx-users] handling particle decomposition with distance restraints
Yes my distance restraints are long because I am using FRET distances as distance restraints while performing MD simulations. Upon usage of this command mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro -pd I get the following error!! I did try giving yes after -pd but even then the same error message is repeated. Back Off! I just backed up md.log to ./#md.log.9# Reading file pr.tpr, VERSION 4.0 (single precision) NNODES=4, MYRANK=1, HOSTNAME=localhost.localdomain NODEID=1 argc=13 NODEID=3 argc=13 [localhost:17514] *** Process received signal *** [localhost:17514] Signal: Segmentation fault (11) [localhost:17514] Signal code: Address not mapped (1) [localhost:17514] Failing at address: 0x134 [localhost:17514] [ 0] /lib64/libpthread.so.0 [0x38dec0f0f0] [localhost:17514] [ 1] /lib64/libc.so.6(memcpy+0x15b) [0x38de08432b] [localhost:17514] [ 2] /usr/lib64/openmpi/1.2.4-gcc/libmpi.so.0(ompi_convertor_pack+0x152) [0x3886e45392] [localhost:17514] [ 3] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_btl_sm.so(mca_btl_sm_prepare_src+0x13d) [0x7f39dd118a4d] [localhost:17514] [ 4] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_pml_ob1.so(mca_pml_ob1_send_request_start_rndv+0x140) [0x7f39dd735230] [localhost:17514] [ 5] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_pml_ob1.so(mca_pml_ob1_send+0x748) [0x7f39dd72e508] [localhost:17514] [ 6] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_coll_tuned.so(ompi_coll_tuned_bcast_intra_split_bintree+0x91c) [0x7f39dc6f735c] [localhost:17514] [ 7] /usr/lib64/openmpi/1.2.4-gcc/libmpi.so.0(MPI_Bcast+0x15c) [0x3886e4c40c] [localhost:17514] [ 8] mdrun_mpi(bcast_state+0x26c) [0x56d59c] [localhost:17514] [ 9] mdrun_mpi(mdrunner+0x1067) [0x42b807] [localhost:17514] [10] mdrun_mpi(main+0x3b4) [0x431c34] [localhost:17514] [11] /lib64/libc.so.6(__libc_start_main+0xe6) [0x38de01e576] [localhost:17514] [12] mdrun_mpi [0x413339] [localhost:17514] *** End of error message *** mpirun noticed that job rank 0 with PID 17514 on node localhost.localdomain exited on signal 11 (Segmentation fault). 3 additional processes aborted (not shown) On Wed, Jun 24, 2009 at 6:04 PM, Justin A. Lemkul jalem...@vt.edu wrote: jayant james wrote: I just replaced the old gmx 4.0 version with the 4.0.5 version and still the same problem NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) Well, does it work with -pd? It looks like your distance restraints are indeed quite long, so this looks like it is your only option. -Justin WARNING: Can not write distance restraint data to energy file with domain decomposition On Wed, Jun 24, 2009 at 5:10 PM, Justin A. Lemkul jalem...@vt.edumailto: jalem...@vt.edu wrote: jayant james wrote: Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Upgrade to the latest version (4.0.5), since there have been numerous improvements to domain decomposition throughout the development of version 4.0. -Justin Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) WARNING: Can not write distance restraint data to energy file with domain decomposition --- Program mdrun_mpi, VERSION 4.0.2 Source code file: domdec.c, line: 5842Fatal error: There is no domain decomposition for 4 nodes that is compatible with the given box and a minimum cell size of 9.85926 nm Change the number of nodes or mdrun option -rdd or -dds Look in the log file for details on the domain decomposition
Re: [gmx-users] handling particle decomposition with distance restraints
jayant james wrote: Yes my distance restraints are long because I am using FRET distances as distance restraints while performing MD simulations. Upon usage of this command mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro -pd I get the following error!! I did try giving yes after -pd but even then the same error message is repeated. Anything else printed to the screen or log file? LINCS warnings or anything else? Did energy minimization complete successfully? -Justin Back Off! I just backed up md.log to ./#md.log.9# Reading file pr.tpr, VERSION 4.0 (single precision) NNODES=4, MYRANK=1, HOSTNAME=localhost.localdomain NODEID=1 argc=13 NODEID=3 argc=13 [localhost:17514] *** Process received signal *** [localhost:17514] Signal: Segmentation fault (11) [localhost:17514] Signal code: Address not mapped (1) [localhost:17514] Failing at address: 0x134 [localhost:17514] [ 0] /lib64/libpthread.so.0 [0x38dec0f0f0] [localhost:17514] [ 1] /lib64/libc.so.6(memcpy+0x15b) [0x38de08432b] [localhost:17514] [ 2] /usr/lib64/openmpi/1.2.4-gcc/libmpi.so.0(ompi_convertor_pack+0x152) [0x3886e45392] [localhost:17514] [ 3] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_btl_sm.so(mca_btl_sm_prepare_src+0x13d) [0x7f39dd118a4d] [localhost:17514] [ 4] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_pml_ob1.so(mca_pml_ob1_send_request_start_rndv+0x140) [0x7f39dd735230] [localhost:17514] [ 5] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_pml_ob1.so(mca_pml_ob1_send+0x748) [0x7f39dd72e508] [localhost:17514] [ 6] /usr/lib64/openmpi/1.2.4-gcc/openmpi/mca_coll_tuned.so(ompi_coll_tuned_bcast_intra_split_bintree+0x91c) [0x7f39dc6f735c] [localhost:17514] [ 7] /usr/lib64/openmpi/1.2.4-gcc/libmpi.so.0(MPI_Bcast+0x15c) [0x3886e4c40c] [localhost:17514] [ 8] mdrun_mpi(bcast_state+0x26c) [0x56d59c] [localhost:17514] [ 9] mdrun_mpi(mdrunner+0x1067) [0x42b807] [localhost:17514] [10] mdrun_mpi(main+0x3b4) [0x431c34] [localhost:17514] [11] /lib64/libc.so.6(__libc_start_main+0xe6) [0x38de01e576] [localhost:17514] [12] mdrun_mpi [0x413339] [localhost:17514] *** End of error message *** mpirun noticed that job rank 0 with PID 17514 on node localhost.localdomain exited on signal 11 (Segmentation fault). 3 additional processes aborted (not shown) On Wed, Jun 24, 2009 at 6:04 PM, Justin A. Lemkul jalem...@vt.edu mailto:jalem...@vt.edu wrote: jayant james wrote: I just replaced the old gmx 4.0 version with the 4.0.5 version and still the same problem NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd) Well, does it work with -pd? It looks like your distance restraints are indeed quite long, so this looks like it is your only option. -Justin WARNING: Can not write distance restraint data to energy file with domain decomposition On Wed, Jun 24, 2009 at 5:10 PM, Justin A. Lemkul jalem...@vt.edu mailto:jalem...@vt.edu mailto:jalem...@vt.edu mailto:jalem...@vt.edu wrote: jayant james wrote: Hi! I am performing an mpi MD (on a quad core system) run with distance restraints. When I execute this command below without position restraints the MD run is distributed over 4 nodes perfectly well. But when I incorporate the distance restraints I hit a road block mpirun -np 4 mdrun_mpi -s pr -e pr -g md -o traj.trr -c pr.gro I get this error message (below). My pr.mdp and distance restraints files are given below the error message . * Question.* How do I handle this sitation? Do I increase the long range cut off in the pr.mdp file? If you see my distance restraints file, my upper range of distances are close to 9nm!! Upgrade to the latest version (4.0.5), since there have been numerous improvements to domain decomposition throughout the development of version 4.0. -Justin Please guide. Thanks JJ -- Back Off! I just backed up md.log to ./#md.log.6# Reading file pr.tpr, VERSION 4.0 (single precision) NOTE: atoms involved in distance restraints should be within the longest cut-off distance, if this is not the case mdrun generates a fatal error, in that case use particle decomposition (mdrun option -pd)