2 # Environment Variables
4 GROMACS programs may be influenced by the use of
5 environment variables. First of all, the variables set in
6 the `GMXRC` file are essential for running and
7 compiling GROMACS. Some other useful environment variables are
8 listed in the following sections. Most environment variables function
9 by being set in your shell to any non-NULL value. Specific
10 requirements are described below if other values need to be set. You
11 should consult the documentation for your shell for instructions on
12 how to set environment variables in the current shell, or in config
13 files for future shells. Note that requirements for exporting
14 environment variables to jobs run under batch control systems vary and
15 you should consult your local documentation for details.
19 * `GMX_CONSTRAINTVIR`: print constraint virial and force virial energy terms.
21 * `GMX_MAXBACKUP`: GROMACS automatically backs up old
22 copies of files when trying to write a new file of the same
23 name, and this variable controls the maximum number of
24 backups that will be made, default 99. If set to 0 it fails to
25 run if any output file already exists. And if set to -1 it
26 overwrites any output file without making a backup.
28 * `GMX_NO_QUOTES`: if this is explicitly set, no cool quotes
29 will be printed at the end of a program.
31 * `GMX_SUPPRESS_DUMP`: prevent dumping of step files during
32 (for example) blowing up during failure of constraint
35 * `GMX_TPI_DUMP`: dump all configurations to a [.pdb]
36 file that have an interaction energy less than the value set
37 in this environment variable.
39 * `GMX_VIEW_XPM`: `GMX_VIEW_XVG`, `GMX_VIEW_EPS` and `GMX_VIEW_PDB`, commands used to
40 automatically view [.xvg], [.xpm], [.eps]
41 and [.pdb] file types, respectively; they default to `xv`, `xmgrace`,
42 `ghostview` and `rasmol`. Set to empty to disable
43 automatic viewing of a particular file type. The command will
44 be forked off and run in the background at the same priority
45 as the GROMACS tool (which might not be what you want).
46 Be careful not to use a command which blocks the terminal
47 (e.g. `vi`), since multiple instances might be run.
49 * `GMX_VIRIAL_TEMPERATURE`: print virial temperature energy term
51 * `GMX_LOG_BUFFER`: the size of the buffer for file I/O. When set
52 to 0, all file I/O will be unbuffered and therefore very slow.
53 This can be handy for debugging purposes, because it ensures
54 that all files are always totally up-to-date.
56 * `GMX_LOGO_COLOR`: set display color for logo in [gmx view].
58 * `GMX_PRINT_LONGFORMAT`: use long float format when printing
61 * `GMX_COMPELDUMP`: Applies for computational electrophysiology setups
62 only (see reference manual). The initial structure gets dumped to
63 [.pdb] file, which allows to check whether multimeric channels have
64 the correct PBC representation.
68 * `GMX_PRINT_DEBUG_LINES`: when set, print debugging info on line numbers.
70 * `GMX_DD_NST_DUMP`: number of steps that elapse between dumping
71 the current DD to a PDB file (default 0). This only takes effect
72 during domain decomposition, so it should typically be
73 0 (never), 1 (every DD phase) or a multiple of `nstlist`.
75 * `GMX_DD_NST_DUMP_GRID`: number of steps that elapse between dumping
76 the current DD grid to a PDB file (default 0). This only takes effect
77 during domain decomposition, so it should typically be
78 0 (never), 1 (every DD phase) or a multiple of `nstlist`.
80 * `GMX_DD_DEBUG`: general debugging trigger for every domain
81 decomposition (default 0, meaning off). Currently only checks
82 global-local atom index mapping for consistency.
84 * `GMX_DD_NPULSE`: over-ride the number of DD pulses used
85 (default 0, meaning no over-ride). Normally 1 or 2.
87 * There are a number of extra environment variables like these
88 that are used in debugging - check the code!
90 ## Performance and Run Control
92 * `GMX_DO_GALACTIC_DYNAMICS`: planetary simulations are made possible (just for fun) by setting
93 this environment variable, which allows setting `epsilon_r = -1` in the [.mdp]
94 file. Normally, `epsilon_r` must be greater than zero to prevent a fatal error.
95 See [wwwpage] for example input files for a planetary simulation.
97 * `GMX_ALLOW_CPT_MISMATCH`: when set, runs will not exit if the
98 ensemble set in the [.tpr] file does not match that of the
101 * `GMX_CUDA_NB_EWALD_TWINCUT`: force the use of twin-range cutoff kernel even if `rvdw` =
102 `rcoulomb` after PP-PME load balancing. The switch to twin-range kernels is automated,
103 so this variable should be used only for benchmarking.
105 * `GMX_CUDA_NB_ANA_EWALD`: force the use of analytical Ewald kernels. Should be used only for benchmarking.
107 * `GMX_CUDA_NB_TAB_EWALD`: force the use of tabulated Ewald kernels. Should be used only for benchmarking.
109 * `GMX_CUDA_STREAMSYNC`: force the use of cudaStreamSynchronize on ECC-enabled GPUs, which leads
110 to performance loss due to a known CUDA driver bug present in API v5.0 NVIDIA drivers (pre-30x.xx).
111 Cannot be set simultaneously with `GMX_NO_CUDA_STREAMSYNC`.
113 * `GMX_CYCLE_ALL`: times all code during runs. Incompatible with threads.
115 * `GMX_CYCLE_BARRIER`: calls MPI_Barrier before each cycle start/stop call.
117 * `GMX_DD_ORDER_ZYX`: build domain decomposition cells in the order
118 (z, y, x) rather than the default (x, y, z).
120 * `GMX_DD_USE_SENDRECV2`: during constraint and vsite communication, use a pair
121 of `MPI_SendRecv` calls instead of two simultaneous non-blocking calls
122 (default 0, meaning off). Might be faster on some MPI implementations.
124 * `GMX_DLB_BASED_ON_FLOPS`: do domain-decomposition dynamic load balancing based on flop count rather than
125 measured time elapsed (default 0, meaning off).
126 This makes the load balancing reproducible, which can be useful for debugging purposes.
127 A value of 1 uses the flops; a value > 1 adds (value - 1)*5% of noise to the flops to increase the imbalance and the scaling.
129 * `GMX_DLB_MAX_BOX_SCALING`: maximum percentage box scaling permitted per domain-decomposition
130 load-balancing step (default 10)
132 * `GMX_DD_RECORD_LOAD`: record DD load statistics for reporting at end of the run (default 1, meaning on)
134 * `GMX_DD_NST_SORT_CHARGE_GROUPS`: number of steps that elapse between re-sorting of the charge
135 groups (default 1). This only takes effect during domain decomposition, so should typically
136 be 0 (never), 1 (to mean at every domain decomposition), or a multiple of `nstlist`.
138 * `GMX_DETAILED_PERF_STATS`: when set, print slightly more detailed performance information
139 to the [.log] file. The resulting output is the way performance summary is reported in versions
140 4.5.x and thus may be useful for anyone using scripts to parse [.log] files or standard output.
142 * `GMX_DISABLE_SIMD_KERNELS`: disables architecture-specific SIMD-optimized (SSE2, SSE4.1, AVX, etc.)
143 non-bonded kernels thus forcing the use of plain C kernels.
145 * `GMX_DISABLE_CUDA_TIMING`: timing of asynchronously executed GPU operations can have a
146 non-negligible overhead with short step times. Disabling timing can improve performance in these cases.
148 * `GMX_DISABLE_GPU_DETECTION`: when set, disables GPU detection even if [mdrun] was compiled
151 * `GMX_DISRE_ENSEMBLE_SIZE`: the number of systems for distance restraint ensemble
152 averaging. Takes an integer value.
154 * `GMX_EMULATE_GPU`: emulate GPU runs by using algorithmically equivalent CPU reference code instead of
155 GPU-accelerated functions. As the CPU code is slow, it is intended to be used only for debugging purposes.
156 The behavior is automatically triggered if non-bonded calculations are turned off using `GMX_NO_NONBONDED`
157 case in which the non-bonded calculations will not be called, but the CPU-GPU transfer will also be skipped.
159 * `GMX_ENX_NO_FATAL`: disable exiting upon encountering a corrupted frame in an [.edr]
160 file, allowing the use of all frames up until the corruption.
162 * `GMX_FORCE_UPDATE`: update forces when invoking `mdrun -rerun`.
164 * `GMX_GPU_ID`: set in the same way as `mdrun -gpu_id`, `GMX_GPU_ID`
165 allows the user to specify different GPU id-s, which can be useful for selecting different
166 devices on different compute nodes in a cluster. Cannot be used in conjunction with `mdrun -gpu_id`.
168 * `GMX_IGNORE_FSYNC_FAILURE_ENV`: allow [mdrun] to continue even if
171 * `GMX_LJCOMB_TOL`: when set to a floating-point value, overrides the default tolerance of
172 1e-5 for force-field floating-point parameters.
174 * `GMX_MAX_MPI_THREADS`: sets the maximum number of MPI-threads that [mdrun]
177 * `GMX_MAXCONSTRWARN`: if set to -1, [mdrun] will
178 not exit if it produces too many LINCS warnings.
180 * `GMX_NB_GENERIC`: use the generic C kernel. Should be set if using
181 the group-based cutoff scheme and also sets `GMX_NO_SOLV_OPT` to be true,
182 thus disabling solvent optimizations as well.
184 * `GMX_NB_MIN_CI`: neighbor list balancing parameter used when running on GPU. Sets the
185 target minimum number pair-lists in order to improve multi-processor load-balance for better
186 performance with small simulation systems. Must be set to a positive integer, the default value
187 is optimized for NVIDIA Fermi and Kepler GPUs, therefore changing it is not necessary for
188 normal usage, but it can be useful on future architectures.
190 * `GMX_NBLISTCG`: use neighbor list and kernels based on charge groups.
192 * `GMX_NBNXN_CYCLE`: when set, print detailed neighbor search cycle counting.
194 * `GMX_NBNXN_EWALD_ANALYTICAL`: force the use of analytical Ewald non-bonded kernels,
195 mutually exclusive of `GMX_NBNXN_EWALD_TABLE`.
197 * `GMX_NBNXN_EWALD_TABLE`: force the use of tabulated Ewald non-bonded kernels,
198 mutually exclusive of `GMX_NBNXN_EWALD_ANALYTICAL`.
200 * `GMX_NBNXN_SIMD_2XNN`: force the use of 2x(N+N) SIMD CPU non-bonded kernels,
201 mutually exclusive of `GMX_NBNXN_SIMD_4XN`.
203 * `GMX_NBNXN_SIMD_4XN`: force the use of 4xN SIMD CPU non-bonded kernels,
204 mutually exclusive of `GMX_NBNXN_SIMD_2XNN`.
206 * `GMX_NO_ALLVSALL`: disables optimized all-vs-all kernels.
208 * `GMX_NO_CART_REORDER`: used in initializing domain decomposition communicators. Rank reordering
209 is default, but can be switched off with this environment variable.
211 * `GMX_NO_CUDA_STREAMSYNC`: the opposite of `GMX_CUDA_STREAMSYNC`. Disables the use of the
212 standard cudaStreamSynchronize-based GPU waiting to improve performance when using CUDA driver API
213 ealier than v5.0 with ECC-enabled GPUs.
215 * `GMX_NO_INT`, `GMX_NO_TERM`, `GMX_NO_USR1`: disable signal handlers for SIGINT,
216 SIGTERM, and SIGUSR1, respectively.
218 * `GMX_NO_NODECOMM`: do not use separate inter- and intra-node communicators.
220 * `GMX_NO_NONBONDED`: skip non-bonded calculations; can be used to estimate the possible
221 performance gain from adding a GPU accelerator to the current hardware setup -- assuming that this is
222 fast enough to complete the non-bonded calculations while the CPU does bonded force and PME computation.
224 * `GMX_NO_PULLVIR`: when set, do not add virial contribution to COM pull forces.
226 * `GMX_NOCHARGEGROUPS`: disables multi-atom charge groups, i.e. each atom
227 in all non-solvent molecules is assigned its own charge group.
229 * `GMX_NOPREDICT`: shell positions are not predicted.
231 * `GMX_NO_SOLV_OPT`: turns off solvent optimizations; automatic if `GMX_NB_GENERIC`
234 * `GMX_NSCELL_NCG`: the ideal number of charge groups per neighbor searching grid cell is hard-coded
235 to a value of 10. Setting this environment variable to any other integer value overrides this hard-coded
238 * `GMX_PME_NTHREADS`: set the number of OpenMP or PME threads (overrides the number guessed by
241 * `GMX_PME_P3M`: use P3M-optimized influence function instead of smooth PME B-spline interpolation.
243 * `GMX_PME_THREAD_DIVISION`: PME thread division in the format "x y z" for all three dimensions. The
244 sum of the threads in each dimension must equal the total number of PME threads (set in
247 * `GMX_PMEONEDD`: if the number of domain decomposition cells is set to 1 for both x and y,
248 decompose PME in one dimension.
250 * `GMX_REQUIRE_SHELL_INIT`: require that shell positions are initiated.
252 * `GMX_REQUIRE_TABLES`: require the use of tabulated Coulombic
253 and van der Waals interactions.
255 * `GMX_SCSIGMA_MIN`: the minimum value for soft-core sigma. **Note** that this value is set
256 using the `sc-sigma` keyword in the [.mdp] file, but this environment variable can be used
257 to reproduce pre-4.5 behavior with respect to this parameter.
259 * `GMX_TPIC_MASSES`: should contain multiple masses used for test particle insertion into a cavity.
260 The center of mass of the last atoms is used for insertion into the cavity.
262 * `GMX_USE_GRAPH`: use graph for bonded interactions.
264 * `GMX_VERLET_BUFFER_RES`: resolution of buffer size in Verlet cutoff scheme. The default value is
265 0.001, but can be overridden with this environment variable.
267 * `MPIRUN`: the `mpirun` command used by [gmx tune_pme].
269 * `MDRUN`: the [mdrun] command used by [gmx tune_pme].
271 * `GMX_NSTLIST`: sets the default value for `nstlist`, preventing it from being tuned during
272 [mdrun] startup when using the Verlet cutoff scheme.
274 * `GMX_USE_TREEREDUCE`: use tree reduction for nbnxn force reduction. Potentially faster for large number of
275 OpenMP threads (if memory locality is important).
277 ## Analysis and Core Functions
279 * `GMX_QM_ACCURACY`: accuracy in Gaussian L510 (MC-SCF) component program.
281 * `GMX_QM_ORCA_BASENAME`: prefix of [.tpr] files, used in Orca calculations
282 for input and output file names.
284 * `GMX_QM_CPMCSCF`: when set to a nonzero value, Gaussian QM calculations will
285 iteratively solve the CP-MCSCF equations.
287 * `GMX_QM_MODIFIED_LINKS_DIR`: location of modified links in Gaussian.
289 * `DSSP`: used by [gmx do_dssp] to point to the `dssp`
290 executable (not just its path).
292 * `GMX_QM_GAUSS_DIR`: directory where Gaussian is installed.
294 * `GMX_QM_GAUSS_EXE`: name of the Gaussian executable.
296 * `GMX_DIPOLE_SPACING`: spacing used by [gmx dipoles].
298 * `GMX_MAXRESRENUM`: sets the maximum number of residues to be renumbered by
299 [gmx grompp]. A value of -1 indicates all residues should be renumbered.
301 * `GMX_FFRTP_TER_RENAME`: Some force fields (like AMBER) use specific names for N- and C-
302 terminal residues (NXXX and CXXX) as [.rtp] entries that are normally renamed. Setting
303 this environment variable disables this renaming.
305 * `GMX_PATH_GZIP`: `gunzip` executable, used by [gmx wham].
307 * `GMX_FONT`: name of X11 font used by [gmx view].
309 * `GMXTIMEUNIT`: the time unit used in output files, can be
310 anything in fs, ps, ns, us, ms, s, m or h.
312 * `GMX_QM_GAUSSIAN_MEMORY`: memory used for Gaussian QM calculation.
314 * `MULTIPROT`: name of the `multiprot` executable, used by the
315 contributed program `do_multiprot`.
317 * `NCPUS`: number of CPUs to be used for Gaussian QM calculation
319 * `GMX_ORCA_PATH`: directory where Orca is installed.
321 * `GMX_QM_SA_STEP`: simulated annealing step size for Gaussian QM calculation.
323 * `GMX_QM_GROUND_STATE`: defines state for Gaussian surface hopping calculation.
325 * `GMX_TOTAL`: name of the `total` executable used by the contributed
328 * `GMX_ENER_VERBOSE`: make [gmx energy] and [gmx eneconv]
331 * `VMD_PLUGIN_PATH`: where to find VMD plug-ins. Needed to be
332 able to read file formats recognized only by a VMD plug-in.
334 * `VMDDIR`: base path of VMD installation.
336 * `GMX_USE_XMGR`: sets viewer to `xmgr` (deprecated) instead of `xmgrace`.