[gmx-users] Implicit solvent crashes with PD

[Mark Abraham]

On 12/07/2011 4:51 AM, Justin A. Lemkul wrote:
>
> Hi All,
>
> I'm having a strange problem with some implicit solvent systems and 
> I'm wondering if anyone's experienced the same problem or if I've 
> stumbled upon a bug.  I'm testing some methodology with a robust 
> system (everyone's favorite, 1AKI lysozyme).  Running simulations with 
> finite cutoffs works fine (in serial or parallel using DD), but if I 
> try to call the all-vs-all kernels (e.g. infinite cutoffs), the 
> simulations instantly crash (at step 0) with LINCS warnings.  Running 
> in serial, however, produces perfectly stable trajectories using 
> all-vs-all, but the simulations are very slow.
>
> The effects are independent of integrator (tested sd and md) and 
> hardware/compilers.  The problem is reproducible on two very different 
> systems:
>
> 1. x86_64 via OpenMPI 1.4.2 and compiled with cmake with gcc-4.3.4
>
> 2. Mac OSX with threads and compiled via autoconf with gcc-4.4.4
>
> All force fields that I've tested (OPLS-AA, CHARMM27, and AMBER03) 
> give the same result.
>
> The problem is present in version 4.5.4 and the most current 
> release-4-5-patches.  My .mdp file and a gdb backtrace are provided at 
> the end of this message.
>
> Is there anything else to try, or should I just file a redmine issue 
> with a .tpr file?

Judging by the .mdp file, I have similar systems working just fine in 
parallel. If you upload the .tpr to Redmine, I'll try it out.

Mark

> -Justin
>
>
> === .mdp file ===
>
> constraints         =  all-bonds
> integrator          =  sd
> dt                  =  0.002    ; ps !
> nsteps              =  50000    ; 100 ps
> nstlist             =  0
> ns_type             =  simple    ; grid also fails
> rlist               =  0
> coulombtype         =  cut-off
> vdwtype             =  cut-off
> rcoulomb            =  0
> rvdw                =  0
> pbc                 =  no
> epsilon_rf          =  0
> rgbradii            =  0
> comm_mode           =  angular
>
> implicit_solvent    = GBSA
> gb_algorithm        = OBC
> gb_epsilon_solvent  = 78.3
> sa_surface_tension  = 2.25936
>
> nstxout             = 0
> nstfout             = 0
> nstvout             = 0
> nstxtcout           = 500
> nstlog              = 500
> nstcalcenergy       = -1
> nstenergy           = 500
>
> tc_grps             = system
> tau_t               = 1.0
> ref_t               = 298
>
> gen_vel             = yes
> gen_temp            = 298
> gen_seed            = 1023
>
>
> === debug output ===
>
> (following the usual LINCS warnings)
>
> (gdb) up
> #1  0x002cf045 in pbc_rvec_sub (pbc=0x0, xi=0x62fae58, xj=0xf954c55c, 
> dx=0xbfffc1cc) at genborn.c:106
> 106            rvec_sub(xi,xj,dx);
> (gdb) up
> #2  0x002d36c0 in gb_bonds_tab (x=0x39000, f=0xdfe000, 
> fshift=0x1033e00, charge=0x10d3e00, p_gbtabscale=0x1032028, 
> invsqrta=0x105e000, dvda=0x1060000, GBtab=0xb35010, idef=0x1092600, 
> epsilon_r=1, gb_epsilon_solvent=78.3000031, facel=137.161057, pbc=0x0, 
> graph=0x0) at genborn.c:1373
> 1373                ki            = pbc_rvec_sub(pbc,x[ai],x[aj],dx);
> (gdb) up
> #3  0x002d4829 in calc_gb_forces (cr=0x903890, md=0x905580, 
> born=0x905370, top=0x1092600, atype=0x1092b1c, x=0x39000, f=0xdfe000, 
> fr=0x1031e00, idef=0x1092600, gb_algorithm=2, sa_algorithm=0, 
> nrnb=0x1031000, bRad=1, pbc=0xbfffc5dc, graph=0x0, enerd=0x9070b0) at 
> genborn.c:1680
> 1680        enerd->term[F_GBPOL]       += gb_bonds_tab(x,f,fr->fshift, 
> md->chargeA,&(fr->gbtabscale),
> (gdb) up
> #4  0x002a4ee1 in do_force_lowlevel (fplog=0xa07d94c0, step=0, 
> fr=0x1031e00, ir=0x1001a00, idef=0x1092600, cr=0x903890, 
> nrnb=0x1031000, wcycle=0x904f30, md=0x905580, opts=0x1001c84, 
> x=0x39000, hist=0x907f50, f=0xdfe000, enerd=0x9070b0, fcd=0x904e10, 
> mtop=0x900c00, top=0x1092600, born=0x905370, atype=0x1092b1c, 
> bBornRadii=1, box=0x907e40, lambda=0, graph=0x0, excl=0x1092b44, 
> mu_tot=0x1031e70, flags=501, cycles_pme=0xbfffc914) at force.c:267
> 267            calc_gb_forces(cr,md,born,top,atype,x,f,fr,idef,
> (gdb) up
> #5  0x003373d5 in do_force (fplog=0xa07d94c0, cr=0x903890, 
> inputrec=0x1001a00, step=0, nrnb=0x1031000, wcycle=0x904f30, 
> top=0x1092600, mtop=0x900c00, groups=0x900c70, box=0x907e40, 
> x=0x39000, hist=0x907f50, f=0xdfe000, vir_force=0xbfffd4d4, 
> mdatoms=0x905580, enerd=0x9070b0, fcd=0x904e10, lambda=0, graph=0x0, 
> fr=0x1031e00, vsite=0x0, mu_tot=0xbfffd434, t=0, field=0x0, ed=0x0, 
> bBornRadii=1, flags=501) at sim_util.c:725
> 725        do_force_lowlevel(fplog,step,fr,inputrec,&(top->idef),
> (gdb) up
> #6  0x000173cc in do_md (fplog=0xa07d94c0, cr=0x903890, nfile=28, 
> fnm=0xbfffed60, oenv=0x900500, bVerbose=1, bCompact=1, 
> nstglobalcomm=10, vsite=0x0, constr=0x905400, stepout=100, 
> ir=0x1001a00, top_global=0x900c00, fcd=0x904e10, 
> state_global=0x900d20, mdatoms=0x905580, nrnb=0x1031000, 
> wcycle=0x904f30, ed=0x0, fr=0x1031e00, repl_ex_nst=0, repl_ex_seed=-1, 
> cpt_period=15, max_hours=-1, deviceOptions=0x24214 "", Flags=7680, 
> runtime=0xbfffd758) at md.c:1015
> 1015                
> do_force(fplog,cr,ir,step,nrnb,wcycle,top,top_global,groups,
> (gdb) up
> #7  0x00013210 in mdrunner (nthreads_requested=2, fplog=0xa07d94c0, 
> cr=0x903890, nfile=28, fnm=0xbfffed60, oenv=0x900500, bVerbose=1, 
> bCompact=1, nstglobalcomm=-1, ddxyz=0xbfffea10, dd_node_order=1, 
> rdd=0, rconstr=0, dddlb_opt=0x2756f "auto", dlb_scale=0.800000012, 
> ddcsx=0x0, ddcsy=0x0, ddcsz=0x0, nstepout=100, resetstep=-1, 
> nmultisim=0, repl_ex_nst=0, repl_ex_seed=-1, pforce=-1, cpt_period=15, 
> max_hours=-1, deviceOptions=0x24214 "", Flags=7680) at runner.c:811
> 811            integrator[inputrec->eI].func(fplog,cr,nfile,fnm,
> (gdb) up
> #8  0x0001cb7b in main (argc=7, argv=0xbffff3c8) at mdrun.c:662
> 662      rc = mdrunner(nthreads, 
> fplog,cr,NFILE,fnm,oenv,bVerbose,bCompact,
>
>