Justin Lemkul [Wed, 19 Jan 2011 16:33:54 +0000 (11:33 -0500)]
Tons of tiny changes to documentation. Manual looks prettier now.
IssueID #671
Justin Lemkul [Wed, 19 Jan 2011 15:00:26 +0000 (10:00 -0500)]
Trivial fixes to program descriptions.
To keep everything in the manual consistent, a few names must
be changed to reflect the new naming scheme that has been adopted.
Justin Lemkul [Tue, 18 Jan 2011 15:21:55 +0000 (10:21 -0500)]
Fixes to the demo script.
Various typographical and editorial changes,
no major changes to functionality.
IssueID #669
Mark Abraham [Tue, 18 Jan 2011 01:36:43 +0000 (12:36 +1100)]
Improved .mdp file comment about the 'define' field
The -D examples were less helpful than the new ones, which provide
a hint not to use -DPOSRE, and a clue that variable names are
case-sensitive.
Justin Lemkul [Mon, 17 Jan 2011 22:16:12 +0000 (17:16 -0500)]
Fixes to methanol and mixed demos. Issue ID #669
Justin Lemkul [Mon, 17 Jan 2011 20:42:44 +0000 (15:42 -0500)]
Removed non-existent program and redundant information to clean up manual
Berk Hess [Mon, 17 Jan 2011 17:21:19 +0000 (18:21 +0100)]
fixed incorrect parameter declarations in gmx_fft_fftw2.c
Justin Lemkul [Mon, 17 Jan 2011 12:50:04 +0000 (07:50 -0500)]
Various editorial and formatting fixes
Mark Abraham [Mon, 17 Jan 2011 09:19:20 +0000 (20:19 +1100)]
Fix Angstrom <=> nm error in documentation
The default value for gb_dielectric_offset .mdp parameter was
listed as 0.09 nm. In the code it is 0.009 nm, which is what it
should be, per the original literature.
Justin Lemkul [Fri, 14 Jan 2011 14:00:18 +0000 (09:00 -0500)]
Fixes to grompp -maxwarn documentation. Issue ID #627
Mark Abraham [Thu, 13 Jan 2011 08:37:58 +0000 (19:37 +1100)]
Made src/kernel/repl_ex.c legible
Some of the formatting was extremely awkward to read, so I
re-formatted the whole file with emacs using the formatting line, and
added braces where consistent with the coding conventions.
IssueID #664 point 2.
Berk Hess [Mon, 10 Jan 2011 09:12:12 +0000 (10:12 +0100)]
fixed zero viscosity energy/log file output with cos_acceleration
Mark Abraham [Fri, 7 Jan 2011 23:12:03 +0000 (10:12 +1100)]
Added note that g_msd -mol does treat PBC (by making molecules whole)
Mark Abraham [Fri, 7 Jan 2011 22:55:00 +0000 (09:55 +1100)]
Implemented effective test for MPI_IN_PLACE, thanks to Michael Wild on the CMake mailing list. Fixes #594
Mark Abraham [Thu, 6 Jan 2011 09:19:56 +0000 (20:19 +1100)]
Stop CMake defaulting to turning off compiler warnings for Intel
compilers, since a) warnings are useful, and b) we show all warnings
with GCC.
ckutzne [Tue, 4 Jan 2011 17:28:12 +0000 (18:28 +0100)]
Fixed inconvenient output file names if -sep is used in combination with -nzero.
For the first file (numbered 0) there was one digit too many,
now all file numbers have nzero digits.
Berk Hess [Tue, 4 Jan 2011 15:18:20 +0000 (16:18 +0100)]
fixed grompp looking for include files in double relative paths, for instance in ../.. with ../topol.top
David van der Spoel [Thu, 23 Dec 2010 08:19:48 +0000 (09:19 +0100)]
Fixed bug in statistics which did not allow fitting with weights.
Added check for return value from statistics codes in g_analyze.
Berk Hess [Wed, 22 Dec 2010 09:44:17 +0000 (10:44 +0100)]
removed incorrect citation print of Engin2010 for pull groups of a single atom
Berk Hess [Tue, 21 Dec 2010 14:38:44 +0000 (15:38 +0100)]
pme.c work-around for MPI fatal error in pme with partticle decomposition
Berk Hess [Thu, 16 Dec 2010 20:28:44 +0000 (21:28 +0100)]
fixed incorrect LJ table in mdrun with reppow!=12
Berk Hess [Thu, 16 Dec 2010 10:41:15 +0000 (11:41 +0100)]
removed editconf warning on box distance with -bt
Berk Hess [Mon, 13 Dec 2010 16:45:35 +0000 (17:45 +0100)]
fixed tpbconv time options leading to negative nsteps for nsteps larger than INT_MAX
David van der Spoel [Sat, 11 Dec 2010 20:35:10 +0000 (21:35 +0100)]
Merge branch 'release-4-5-patches' of git.gromacs.org:gromacs into release-4-5-patches
David van der Spoel [Sat, 11 Dec 2010 20:34:05 +0000 (21:34 +0100)]
Added 1 quote.
Pär Bjelkmar [Fri, 10 Dec 2010 12:22:25 +0000 (13:22 +0100)]
Added CHARMM neutral lysine, LYN, to residuetypes.dat.
Christoph Junghans [Fri, 10 Dec 2010 10:11:30 +0000 (11:11 +0100)]
Fixed wrong path ordering in GMX.bash introduced in commit
58a6b0bee590c2293fbf3eb35947005db9f7668b
David van der Spoel [Thu, 9 Dec 2010 11:26:42 +0000 (12:26 +0100)]
Added fatal error for when people use energy files for computing dipoles. This
is broken due to the fact that the dipole is computed based on broken
molecules.
Mark Abraham [Mon, 6 Dec 2010 08:45:32 +0000 (19:45 +1100)]
Added comments and workaround for IBM's BlueGene(/L) MPI_Bcast dereferencing
null pointers even when zero data is to be broadcast. This change will only affect
BlueGene, and does not affect speed of time-critical parts of GROMACS.
ckutzne [Thu, 2 Dec 2010 14:44:41 +0000 (15:44 +0100)]
Fixed typo.
Christoph Junghans [Mon, 29 Nov 2010 11:35:18 +0000 (12:35 +0100)]
GMXRC.* - fixed bug #614
Christoph Junghans [Fri, 26 Nov 2010 09:59:54 +0000 (10:59 +0100)]
added power6 flags from Markus Rampp
David van der Spoel [Sun, 28 Nov 2010 10:11:57 +0000 (11:11 +0100)]
Fixed bugzilla 616 about formatting in the help text.
David van der Spoel [Fri, 26 Nov 2010 10:10:18 +0000 (11:10 +0100)]
Fixed SEGV bug in g_sham.
Berk Hess [Wed, 24 Nov 2010 12:44:49 +0000 (13:44 +0100)]
fixed segv in grompp due to gmxcpp not checking for #include statements without delimiters
Christoph Junghans [Wed, 24 Nov 2010 08:53:29 +0000 (09:53 +0100)]
configure: build shared libs unless mpi enabled (bug #562)
Mark Abraham [Mon, 22 Nov 2010 12:07:47 +0000 (23:07 +1100)]
Added a hint to the documentation for creating .tpr files with a subset of a simulation sustem using tpbconv.
Sander Pronk [Fri, 19 Nov 2010 13:38:44 +0000 (14:38 +0100)]
Fixed TMPI_WAIT_FOR_NO_ONE deadlock.
David van der Spoel [Thu, 18 Nov 2010 19:45:00 +0000 (20:45 +0100)]
Made printing in check_multi_init conditional, depending on whether the
log file is NULL or not.
Mark Abraham [Fri, 12 Nov 2010 23:03:20 +0000 (10:03 +1100)]
Minor trjconv documentation formatting fix
Mark Abraham [Fri, 12 Nov 2010 23:01:52 +0000 (10:01 +1100)]
g_densmap -od writes a .dat file and prevents writing the .xpm file. The latter behaviour is uncharacteristic of GROMACS tools, I thought, so I updated the documentation to note this. Perhaps the .xpm file produced by -o should always be written.
Rossen Apostolov [Tue, 9 Nov 2010 21:32:11 +0000 (22:32 +0100)]
New maintanance release: 4.5.3
Berk Hess [Tue, 9 Nov 2010 08:47:50 +0000 (09:47 +0100)]
added -resnr option to editconf
Mark Abraham [Tue, 9 Nov 2010 08:16:06 +0000 (19:16 +1100)]
GMX_POWERPC_INVSQRT should be OFF in CMakeLists.txt by default (PowerPC and BlueGene accelerations turn them on, though)
Berk Hess [Tue, 9 Nov 2010 07:49:18 +0000 (08:49 +0100)]
fixed compiler warning in pdb2gmx.c
Rossen Apostolov [Mon, 8 Nov 2010 19:16:09 +0000 (20:16 +0100)]
Reverting back GMX_DLOPEN to default ON
Roland Schulz [Mon, 8 Nov 2010 18:57:01 +0000 (13:57 -0500)]
making error message for vmdio clearer (bug 598)
Pär Bjelkmar [Mon, 8 Nov 2010 16:17:42 +0000 (17:17 +0100)]
Deleted redundant charmm27.ff/dna.r2b.
Rossen Apostolov [Mon, 8 Nov 2010 15:55:22 +0000 (16:55 +0100)]
Disable GMX_DLOPEN by default.
Bug 598 is still open though.
Rossen Apostolov [Mon, 8 Nov 2010 15:45:56 +0000 (16:45 +0100)]
Added support for Altivec kernels on POWERPC.
Hopefully closes bug 573.
Szilard Pall [Mon, 8 Nov 2010 12:49:29 +0000 (13:49 +0100)]
Merge branch 'release-4-5-patches' of git.gromacs.org:gromacs into release-4-5-patches
Mark Abraham [Sun, 7 Nov 2010 01:36:37 +0000 (12:36 +1100)]
Made mdrun exit with a fatal error for EM or NM with -rerun
Mark Abraham [Sat, 6 Nov 2010 11:55:56 +0000 (22:55 +1100)]
Fixed .xvg legends for g_sas options -or and -oa
Szilard Pall [Fri, 5 Nov 2010 22:08:50 +0000 (23:08 +0100)]
Merge branch 'release-4-5-patches' of git.gromacs.org:gromacs into release-4-5-patches
Pär Bjelkmar [Fri, 5 Nov 2010 12:38:11 +0000 (13:38 +0100)]
More atom-based charge groups in CHARMM (termini and hard-coded residues).
Berk Hess [Fri, 5 Nov 2010 09:18:29 +0000 (10:18 +0100)]
pdb2gmx now puts atoms with cgnr=-1 in separate charge groups
Roland Schulz [Thu, 4 Nov 2010 08:14:56 +0000 (04:14 -0400)]
print more information in case of lock error
Berk Hess [Thu, 4 Nov 2010 10:19:35 +0000 (11:19 +0100)]
Fixed md.log being opened twice when appending, which prevented file locking.
Added check for -seppot and -append.
hess [Thu, 4 Nov 2010 08:42:54 +0000 (09:42 +0100)]
added GB params for charmm MCH3S
hess [Thu, 4 Nov 2010 08:42:18 +0000 (09:42 +0100)]
clarified GB parameter error messages in grompp
Szilard Pall [Wed, 3 Nov 2010 11:32:52 +0000 (12:32 +0100)]
now it's possible to run binaries in the build tree (fixes bug #570)
Changed CMAKE_BUILD_WITH_INSTALL_RPATH to FALSE which results in the
rpath being rewritten on while installing binaries. CPack also works.
Berk Hess [Tue, 2 Nov 2010 17:29:26 +0000 (18:29 +0100)]
fixed incorrect print in fatal error on charge group size
Berk Hess [Tue, 2 Nov 2010 17:16:13 +0000 (18:16 +0100)]
proper defined of MAX_CHARGEGROUP_SIZE=32 in a single include file
David van der Spoel [Tue, 2 Nov 2010 15:09:35 +0000 (16:09 +0100)]
Fixed bug 608 by throwing a fatal error in grompp.
Sander Pronk [Tue, 2 Nov 2010 12:27:01 +0000 (13:27 +0100)]
Fixed double precision energy file reading + box matrix writing to ener.edr
- double precision energy files can now be read back
- All non-zero elements of the box matrix are now correctly written out
when needed
hess [Tue, 2 Nov 2010 07:50:33 +0000 (08:50 +0100)]
fixed incorrect y result in mvmul_ur0, only used in g_vanhove
Rossen Apostolov [Sat, 30 Oct 2010 20:07:36 +0000 (22:07 +0200)]
Added back the "-dev" suffixes to the version strings.
Rossen Apostolov [Sat, 30 Oct 2010 19:22:54 +0000 (21:22 +0200)]
New bugfix release: 4.5.2
Rossen Apostolov [Sat, 30 Oct 2010 17:27:54 +0000 (19:27 +0200)]
For MPI builds, CMake will first try to compile a sample MPI program.
On some systems such as Cray the FindMPI.cmake module can't find the proper
libraries to link to. Therefore we now test whether the given C compiler
is an MPI wrapper and skip the rest of the tests.
Thanks to Roland Schulz for the patch.
Sander Pronk [Sat, 30 Oct 2010 15:11:18 +0000 (17:11 +0200)]
Added Windows macro for snprintf.
Rossen Apostolov [Fri, 29 Oct 2010 19:24:17 +0000 (21:24 +0200)]
added g_kinetics to the CMake build scripts
Szilard Pall [Fri, 29 Oct 2010 17:06:10 +0000 (19:06 +0200)]
Merge branch 'release-4-5-patches' of git.gromacs.org:gromacs into release-4-5-patches
Szilard Pall [Fri, 29 Oct 2010 17:04:44 +0000 (19:04 +0200)]
enabled GMX_PREFER_STATIC_LIBS CMake option for CYGWIN
As it turns out that Cygwin also uses ".a" for static libraries,
the it is safe to enable this option.
Carsten Kutzner [Fri, 29 Oct 2010 16:38:33 +0000 (18:38 +0200)]
Fixed segv in gmx_hbond
Szilard Pall [Fri, 29 Oct 2010 15:24:06 +0000 (17:24 +0200)]
fixed the default for OpenMM plugin dir location (bug #606)
The OpenMM plugin lookup mechanism works again, so when running mdrun-gpu
on the same machine/OS where the binaries were built, the plugins are picked
up from the OpenMM installation used for building (but the env.var.
OPENMM_PLUGIN_DIR still has priority over all defaults!).
Szilard Pall [Fri, 29 Oct 2010 15:28:02 +0000 (17:28 +0200)]
fixed make install-mdrun issue (bug #605)
Szilard Pall [Fri, 29 Oct 2010 14:18:01 +0000 (16:18 +0200)]
(re)added "-dev" suffix for the CMake hardcoded version number
This avoids the potential confusion when the source from a git
release banch is used on machines without git. In these cases the
version reported by gromacs binaries would be the same as the
version of an official release, while former might actually
differ from the release in some patches from git.
Szilard Pall [Fri, 29 Oct 2010 13:44:31 +0000 (15:44 +0200)]
added CMake option GMX_PREFER_STATIC_LIBS (also fixes bug #541)
This option instructs CMake to prefer static external libs when both static
and shared version available. This makes possible linking statically against
external libraries (e.g. fftw, xml2).
The option only available on *NIX platforms, except APPLE and CYGWIN.
Note, that when the option is enabled UILD_SHARED_LIBS gets disabled.
Also note, that with this option on, the binaries built will still be
dynamically linked against system libraries.
David van der Spoel [Thu, 28 Oct 2010 20:26:28 +0000 (22:26 +0200)]
Merge branch 'release-4-5-patches' of git.gromacs.org:gromacs into release-4-5-patches
Per Larsson [Thu, 28 Oct 2010 18:44:59 +0000 (20:44 +0200)]
Fixed a bug in the all-vs-all sse gb kernels, where an exclusion mask was
calculated incorrectly. Also fixes to loop indeces in the all-vs-all non-sse
code.
Berk Hess [Thu, 28 Oct 2010 10:12:24 +0000 (12:12 +0200)]
added HEM to HEME renaming for Gromos and Charmm ff's
Berk Hess [Thu, 28 Oct 2010 10:09:13 +0000 (12:09 +0200)]
fixed pdb2gmx chain merging bug (bugzilla 544)
hess [Wed, 27 Oct 2010 21:23:58 +0000 (23:23 +0200)]
changed fftw2_empty to fftw3_empty in gmx_fft_fftw3.c
Berk Hess [Tue, 26 Oct 2010 09:04:41 +0000 (11:04 +0200)]
fixed cycle print type cast issue on certain platforms, fixes bugzilla 601
Mark Abraham [Sun, 24 Oct 2010 16:19:10 +0000 (03:19 +1100)]
Don't sprintf variable-sized input into fixed-size buffers
Mark Abraham [Fri, 22 Oct 2010 16:01:00 +0000 (05:01 +1300)]
Re-enabled for BlueGene/L an optimization in the BG kernels that works for BG/L but not always for BG/P. Bugzilla 429 says this optimization is not stable for use in BG/P, however the work-around in
764d171c36b32c0497090531a11de disables it for BG/P too, which is undesirable.
Mark Abraham [Fri, 22 Oct 2010 15:58:34 +0000 (04:58 +1300)]
Rearranged cmake+BlueGene solution to have the same form as the other GMX_ACCELERATION solutions
Mark Abraham [Fri, 22 Oct 2010 15:57:04 +0000 (04:57 +1300)]
Made GMX_POWERPC_INVSQRT work properly on Power6 and BlueGene
Mark Abraham [Fri, 22 Oct 2010 06:02:25 +0000 (19:02 +1300)]
Power6 FORTRAN kernels do not work on PPC440 or PPC450 (these are the BlueGene chips)
Mark Abraham [Fri, 22 Oct 2010 05:50:47 +0000 (18:50 +1300)]
Carsten Kutzner [Fri, 22 Oct 2010 09:51:32 +0000 (11:51 +0200)]
Fixed segv in g_potential. Field output has now also the correct prefactor for V/nm
Per Larsson [Fri, 22 Oct 2010 08:31:02 +0000 (10:31 +0200)]
Added a check in grompp for the case with particles having a non-zero
charge while all GB-parameters are set to zero. This fixes bug #599.
Berk Hess [Thu, 21 Oct 2010 07:45:44 +0000 (09:45 +0200)]
clarified make_ndx error message
Mark Abraham [Wed, 20 Oct 2010 13:16:31 +0000 (02:16 +1300)]
Fixed cmake to work with BlueGene/L. This fix does not work for BlueGene/P.
Sander Pronk [Wed, 20 Oct 2010 13:00:57 +0000 (15:00 +0200)]
file reading error messages in g_bar less confusing.
Per Larsson [Wed, 20 Oct 2010 07:07:25 +0000 (09:07 +0200)]
Removed code for calculating logarithms using a table
This was not much faster than the builtin c-function,
and also only used in the non-sse loops
Berk Hess [Tue, 19 Oct 2010 13:36:01 +0000 (15:36 +0200)]
changed enxio warning into fatal errors by default, can be overridden with an env var
Per Larsson [Tue, 19 Oct 2010 08:32:21 +0000 (10:32 +0200)]
Fixed a compiler warning introduced by commit
9ccc786e4...
Per Larsson [Tue, 19 Oct 2010 05:52:58 +0000 (07:52 +0200)]
Replaced _mm_castsi128_pd with gmx_mm_castsi128_pd everywhere
in gmx_sse2_double
Per Larsson [Mon, 18 Oct 2010 19:50:40 +0000 (21:50 +0200)]
Double precision GB energies are now summed corrently (bug #585)
The issue was that __m128d _mm_add_sd(a,b) should give back r0 = a0 + b0
and r1 = a0, but it gives back r1 = 0. But since this is not true for
all architectures, it is also important to carefully zeroing out of some
variables in the code (explained in the kernel4XX-comments).