[gmx-developers] TPI code with Verlet cut-off scheme in GROMACS-5.0

[Berk Hess]

Hi,

Gromacs 5.0 does not support the combination of TPI and the Verlet 
cut-off scheme. Apparently we forgot to add a check for this. We will 
for 5.0.1.
The next major release will support this combination.

Cheers,

Berk

On 07/30/2014 02:13 PM, Alexander Schlaich wrote:
> Dear Gromacs developers,
>
> I would like to use the TPI code in combination with the Verlet cut-off scheme in Gromacs-5.0. However, when rerunning the trajectory with TPI, I receive a segmentation fault which I guess is related to some incompatibility with the charge-group scheme:
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0x2aaab07b0700 (LWP 20235)]
> 0x00002aaaab2a65ef in copy_rvec (a=0x2aaaaabf4014, b=0x20004) at /home/aschlaich/software/src/gromacs-5/src/gromacs/legacyheaders/vec.h:327
> 327	    b[XX] = a[XX];
> (gdb) bt
> #0  0x00002aaaab2a65ef in copy_rvec (a=0x2aaaaabf4014, b=0x20004) at /home/aschlaich/software/src/gromacs-5/src/gromacs/legacyheaders/vec.h:327
> #1  0x00002aaaab2a6e22 in calc_cgcm (fplog=0x0, cg0=10923, cg1=10924, cgs=0x2aaabc05bc40, pos=0x2aaaaabd4010, cg_cm=0x0) at /home/aschlaich/software/src/gromacs-5/src/gromacs/gmxlib/chargegroup.c:171
> #2  0x00002aaaac41ab2d in do_tpi (fplog=0x0, cr=0x68e3d0, nfile=35, fnm=0x68cc70, oenv=0x6529d0, bVerbose=0, bCompact=1, nstglobalcomm=-1, vsite=0x0, constr=0x2aaabc05af90, stepout=100, inputrec=0x68e450, top_global=0x68e8b0,
>      fcd=0x2aaab8019950, state=0x68eab0, mdatoms=0x6e7fa0, nrnb=0x2aaab4027740, wcycle=0x2aaab8019ac0, ed=0x0, fr=0x2aaab801a800, repl_ex_nst=0, repl_ex_nex=0, repl_ex_seed=-1, membed=0x0, cpt_period=15, max_hours=-1,
>      deviceOptions=0x4358c6 "", imdport=0, Flags=1055760, walltime_accounting=0x2aaabc05af50) at /home/aschlaich/software/src/gromacs-5/src/gromacs/mdlib/tpi.c:288
> #3  0x000000000042cf20 in mdrunner (hw_opt=0x2aaab07afc90, fplog=0x0, cr=0x68e3d0, nfile=35, fnm=0x68cc70, oenv=0x6529d0, bVerbose=0, bCompact=1, nstglobalcomm=-1, ddxyz=0x2aaab07afcfc, dd_node_order=1, rdd=0, rconstr=0,
>      dddlb_opt=0x4358a3 "auto", dlb_scale=0.800000012, ddcsx=0x0, ddcsy=0x0, ddcsz=0x0, nbpu_opt=0x4358a3 "auto", nstlist_cmdline=0, nsteps_cmdline=-2, nstepout=100, resetstep=-1, nmultisim=0, repl_ex_nst=0, repl_ex_nex=0,
>      repl_ex_seed=-1, pforce=-1, cpt_period=15, max_hours=-1, deviceOptions=0x4358c6 "", imdport=0, Flags=1055760) at /home/aschlaich/software/src/gromacs-5/src/programs/mdrun/runner.c:1774
> #4  0x0000000000429155 in mdrunner_start_fn (arg=0x651540) at /home/aschlaich/software/src/gromacs-5/src/programs/mdrun/runner.c:180
> #5  0x00002aaaac4ed273 in tMPI_Thread_starter (arg=0x670c18) at /home/aschlaich/software/src/gromacs-5/src/external/thread_mpi/src/tmpi_init.c:397
> #6  0x00002aaaac4e40cd in tMPI_Thread_starter (arg=0x2aaab4000a10) at /home/aschlaich/software/src/gromacs-5/src/external/thread_mpi/src/pthreads.c:231
> #7  0x00002aaaae4c9b50 in start_thread (arg=<optimized out>) at pthread_create.c:304
> #8  0x00002aaaaf1680ed in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
> #9  0x0000000000000000 in ?? ()
>
> Could you tell me what I would do to fix this if I'm right with my assumption? Do you plan to switch the TPI-code to the Verlet scheme?
>
> Thank you very much for your answer,
>
> Alex
>
>