alexxy/gromacs.git
11 years agoMerge "added mdrun check for Verlet+perturbed atoms" into release-4-6
David van der Spoel [Sat, 13 Apr 2013 14:53:10 +0000 (16:53 +0200)]
Merge "added mdrun check for Verlet+perturbed atoms" into release-4-6

11 years agoMerge "removed x86 specifics from nbnxn SIMD kernels" into release-4-6
Mark Abraham [Fri, 12 Apr 2013 11:46:50 +0000 (13:46 +0200)]
Merge "removed x86 specifics from nbnxn SIMD kernels" into release-4-6

11 years agoMerge "corrected log file wall time breakdown" into release-4-6
Mark Abraham [Fri, 12 Apr 2013 07:55:19 +0000 (09:55 +0200)]
Merge "corrected log file wall time breakdown" into release-4-6

11 years agoMerge "made PME work with a mix of 1 and more threads" into release-4-6
Mark Abraham [Fri, 12 Apr 2013 07:54:08 +0000 (09:54 +0200)]
Merge "made PME work with a mix of 1 and more threads" into release-4-6

11 years agoMerge "Merge release-4-5-patches into release-4-6" into release-4-6
Teemu Murtola [Wed, 10 Apr 2013 17:07:57 +0000 (19:07 +0200)]
Merge "Merge release-4-5-patches into release-4-6" into release-4-6

11 years agoMerge "Added install guide section for BLAS/LAPACK" into release-4-6
Christoph Junghans [Fri, 29 Mar 2013 17:16:37 +0000 (18:16 +0100)]
Merge "Added install guide section for BLAS/LAPACK" into release-4-6

11 years agoremoved x86 specifics from nbnxn SIMD kernels
Berk Hess [Mon, 18 Feb 2013 16:57:45 +0000 (17:57 +0100)]
removed x86 specifics from nbnxn SIMD kernels

All x86 specific preprossing and SIMD intrinsics in
nbxn_kernel_simd_???_outer/inner.h have been replaced by generic SIMD
macro code. Exclusion masks are now loaded iso set and the treatment
is the same for most setups and can use only floating point intrinsics.
Moved half-width SIMD from gmx_simd_macros.h to two 2xnn .h files.
Half-width load macros now take the output as an argument,
in preparation for Intel MIC SIMD instrinsics.
Replaced _SSE suffixes by _S in the abovementioned files.

Change-Id: Ib634726af6e51a99c904e05528b0ac10ecbc2462

11 years agocorrected log file wall time breakdown
Berk Hess [Wed, 27 Mar 2013 09:19:41 +0000 (10:19 +0100)]
corrected log file wall time breakdown

The recent commit 873b9854 doubled the wall times in the breakdown
table in the log file when not using separate PME nodes.
Note that this didn't affect cycle count or total timing prints.
Fixes #1210

Change-Id: Ia2b1d5e6c21543990465b7c7a4a8a1e687c41528

11 years agoMerge release-4-5-patches into release-4-6
Roland Schulz [Tue, 26 Mar 2013 19:15:20 +0000 (15:15 -0400)]
Merge release-4-5-patches into release-4-6

Conflicts:
CMakeLists.txt
configure.ac
src/tools/gmx_mindist.c

Change-Id: I303318f17527a8e1bfe494eb174565b8d5491870

11 years agomade PME work with a mix of 1 and more threads
Berk Hess [Thu, 21 Mar 2013 09:33:07 +0000 (10:33 +0100)]
made PME work with a mix of 1 and more threads

Using a mix of 1 and more OpenMP threads on different MPI ranks
would make mdrun terminate with an MPI error.
Fixes #1171

Change-Id: Iffa16e18baf0f74be826b59503208dca01d1ec14

11 years agoMerge "Essential dynamics: fixed legend in ED .xvg output file" into release-4-6
Christoph Junghans [Tue, 19 Mar 2013 17:42:28 +0000 (18:42 +0100)]
Merge "Essential dynamics: fixed legend in ED .xvg output file" into release-4-6

11 years agoMerge "fix broken things in mkhtml" into release-4-6
Mark Abraham [Tue, 12 Mar 2013 17:36:44 +0000 (18:36 +0100)]
Merge "fix broken things in mkhtml" into release-4-6

11 years agoAdded install guide section for BLAS/LAPACK
Mark Abraham [Mon, 11 Mar 2013 13:15:21 +0000 (14:15 +0100)]
Added install guide section for BLAS/LAPACK

Fixes #1186

Change-Id: If9e95f93cd216b86143685fd0ae9bfbc5d26792e

11 years agoMerge "Fixes linkage with FFTW + MKL for BLAS/LAPACK" into release-4-6
Mark Abraham [Mon, 11 Mar 2013 11:10:49 +0000 (12:10 +0100)]
Merge "Fixes linkage with FFTW + MKL for BLAS/LAPACK" into release-4-6

11 years agoEssential dynamics: fixed legend in ED .xvg output file
Carsten Kutzner [Fri, 8 Mar 2013 16:07:26 +0000 (17:07 +0100)]
Essential dynamics: fixed legend in ED .xvg output file

If one performs flooding only and none of the standard essential dynamics
protocols are turned on, the .xvg output file did still print ED legends.
I moved the bNeedDoEdsam check from init_edi() to write_edo_legend(), so
that already here we can check whether ED or only flooding is performed.

Change-Id: Ic3a784b63b335059c17aed852d1b4a4f3126841a

11 years agoFixes #1183 PBC bug in g_mindist
David van der Spoel [Thu, 7 Mar 2013 18:26:23 +0000 (19:26 +0100)]
Fixes #1183 PBC bug in g_mindist

Fixes minimum size of the box if triclinic when checking
the periodic image distance.

Change-Id: I54cb593c42f791b6540147233c345069f84e2f33

11 years agoMerge "fixed PME time print with -ntomp_pme != -ntomp" into release-4-6
Roland Schulz [Wed, 6 Mar 2013 04:57:14 +0000 (05:57 +0100)]
Merge "fixed PME time print with -ntomp_pme != -ntomp" into release-4-6

11 years agoNew patch release 4.6.1
Mark Abraham [Fri, 15 Feb 2013 23:05:48 +0000 (00:05 +0100)]
New patch release 4.6.1

Change-Id: I60998e3be8c0c7289d0aa5279e554cc79fb74310

11 years agoFix automated regression testing GMXLIB value
Mark Abraham [Tue, 5 Mar 2013 15:56:34 +0000 (16:56 +0100)]
Fix automated regression testing GMXLIB value

Change-Id: I3030a85df4d649106f4620b48374ee41a9bdebc9

11 years agofixed bug with DD cut-off check and PME DLB
Berk Hess [Thu, 28 Feb 2013 17:10:34 +0000 (18:10 +0100)]
fixed bug with DD cut-off check and PME DLB

During the initial stage of PME DLB, the cut-off limit check for the
DD DLB was not checked correcty. This reduced the effectiveness of
both DD and PME DLB and could lead to missing interactions during
a very limited number of steps.
Fixes #1169

Change-Id: I0b0cac22d4abbe8a9b0c16488aeb662d7d4d9bcc

11 years agoFix regressiontests automated download
Mark Abraham [Wed, 27 Feb 2013 15:29:42 +0000 (16:29 +0100)]
Fix regressiontests automated download

Refs #1150

Change-Id: I7ba82b22a75532516f481dcb5e4a870047c0bfe8

11 years agofixed recent bug with sorting atoms for GPUs
Berk Hess [Wed, 20 Feb 2013 09:41:38 +0000 (10:41 +0100)]
fixed recent bug with sorting atoms for GPUs

The sort buffer in the nbnxn gridding for GPUs was made too small
in 8d6cc146. This led to inconsistency errors (not incorrect output).
This is fixed and the sort_atoms call is now made simpler and
multiple consistency checks are now always on in debug builds.
Fixes #1153

Change-Id: Ifcdf45bb4de88e7584628d3ed2699e2fd469d5c6

11 years agoUpdate shell completions
Mark Abraham [Wed, 20 Feb 2013 10:04:48 +0000 (11:04 +0100)]
Update shell completions

There's some strange changes of what completions are being generated
for inputs that take multiple files when generated by different 4.6.x
versions, but nobody seems to know what is going on. Also, those
problems do not have a large impact.

Refs #1159

Change-Id: Ib4b1ad4999e1b68904d46b0d01c2a5dd07507fd7

11 years agoUpdate outdated admin things
Mark Abraham [Fri, 15 Feb 2013 22:54:38 +0000 (23:54 +0100)]
Update outdated admin things

* remove outdated CMake installation guide
* updated the bootstrap script
* remove outdated admin/README.patches
* update README, including adding reference to
  GROMACS 4.5 paper and DOIs for old papers

Change-Id: I9a55126049806ca67f6ad846cdbd67cc530c7db0

11 years agoFixes linkage with FFTW + MKL for BLAS/LAPACK
Roland Schulz [Tue, 5 Mar 2013 09:15:01 +0000 (04:15 -0500)]
Fixes linkage with FFTW + MKL for BLAS/LAPACK

The order of linking was incorrect, when using both mkl and fftw
causing the mkl fftw wrapper to be used instead of fftw itself.

Removes all unnecessary transitive libraries (cmake adds those
automatically). Simplifies the target_link_libraries lines, and
makes it much easier to change the fft/blas link order.

Only remaining transitive libraries are gmx/md for binaries. This
is required because OpenMP_LINKER_FLAGS needs to be at the end
but transitive libraries are added at the end. Thus without listing
gmx/md manually the order is incorrect.

Also fixes that nbnxn_cuda has cuda_tools listed as library
dependency. Was prior working without it but now is needed and is
also anyhow the more corect way to do it.

Fixes #1067

Change-Id: I639b16f1460d27a15eb72cc00674c041266f6624

11 years agoFixes for install guide page.
Justin Lemkul [Fri, 8 Feb 2013 20:40:53 +0000 (15:40 -0500)]
Fixes for install guide page.

Editorial fixes to rename GMX_LIBS_SUFFIX and provide an example
of proper usage of adding suffixes.  These edits now keep the .tex
file in sync with the edits manually made to the wiki page.

Change-Id: Iba5b4c66ad3d0d84ddb06591d4681ff48e4de2ee

11 years agoUpdated install guide
Mark Abraham [Wed, 20 Feb 2013 11:35:46 +0000 (12:35 +0100)]
Updated install guide

* bumped some versions for 4.6.1
* reviewed various promises about 4.6.1

Change-Id: I6d8549e1f47c53367db275c3091b02556af3b8d8

11 years agoWarn if X11 was not found and the user wanted it
Mark Abraham [Tue, 5 Mar 2013 14:02:23 +0000 (15:02 +0100)]
Warn if X11 was not found and the user wanted it

Also, removed unused HAVE_X11 variable.

Change-Id: I581b2036f9de27a02c034230f2d1212b9814ffbb

11 years agofix broken things in mkhtml
Mark Abraham [Tue, 5 Mar 2013 13:45:56 +0000 (14:45 +0100)]
fix broken things in mkhtml

Change-Id: Ifb9aa694eddd4bffe457674265a2b17093182c50

11 years agoMerge "Uncrustified code changes since 4.6" into release-4-6
Christoph Junghans [Sat, 2 Mar 2013 23:01:24 +0000 (00:01 +0100)]
Merge "Uncrustified code changes since 4.6" into release-4-6

11 years agoDeclaration-definition consistency nbxn_cuda_wait_gpu.
Peter Kasson [Thu, 28 Feb 2013 01:32:07 +0000 (17:32 -0800)]
Declaration-definition consistency nbxn_cuda_wait_gpu.

Changed float->real in nbnxn_cuda_wait_gpu() for consistency between
function declaration and function definition.  Added note that this is
only implemented for single-precision at the moment.  CMake will
complain if GMX_GPU and GMX_DOUBLE are both set.
Addresses Bug #1157.

Change-Id: Ide495cbaba6d120d91f106c6a87ca04e46a2f5a8

11 years agofixed PME time print with -ntomp_pme != -ntomp
Berk Hess [Tue, 26 Feb 2013 19:29:54 +0000 (20:29 +0100)]
fixed PME time print with -ntomp_pme != -ntomp

Fixes #1158

Change-Id: I9028e6309b8802125e91935e87d96b341ec30f77

11 years agoUncrustified code changes since 4.6
Mark Abraham [Fri, 15 Feb 2013 23:15:20 +0000 (00:15 +0100)]
Uncrustified code changes since 4.6

The group kernel python scripts have changed slightly to emit
code that uncrustify won't want to change. Group kernel
generation and uncrustify are now mutual null operations.

Recent changes to the declarations in gmx_order in baa65b60
may have exposed a possible bug in uncrustify. Those declarations
have been simplified so that there will be no future issue from
that possible bug.

Refs #845

Change-Id: I33495d57ca37317cef4bf12707fd77d67309d292

11 years agofix imcompatibility with tpr vers<64 with free energy option
Michael Shirts [Mon, 25 Feb 2013 04:03:05 +0000 (23:03 -0500)]
fix imcompatibility with tpr vers<64 with free energy option

should fix to redmine #1156

Change-Id: I6e60f1efbb03bf460009b8bc59f80f2e0e15bf44

11 years agoadded mdrun check for Verlet+perturbed atoms
Berk Hess [Wed, 20 Feb 2013 15:26:12 +0000 (16:26 +0100)]
added mdrun check for Verlet+perturbed atoms

Currently the Verlet scheme does not support perturbed atoms.
This will be implemented soon.

Change-Id: Id8656aa680b6581882f5371514cdbd85f429c418

11 years agoBump shared object version to 8
Christoph Junghans [Fri, 15 Feb 2013 17:29:46 +0000 (10:29 -0700)]
Bump shared object version to 8

libs are not compatible with the ones from Gromacs 4.5.x, which had so
version 6 through 4.5.6, and now so version 7 for 4.5.7.

Fixes #1147

Change-Id: If96fd044b00a99eb4ba8376ffa0a2ecbea37894a

11 years agofixed issues with FEP soft-core and cut-off's
Berk Hess [Fri, 8 Feb 2013 12:43:57 +0000 (13:43 +0100)]
fixed issues with FEP soft-core and cut-off's

Now the cut-off is applied to the PME mesh correction for perturbed
pairs with soft-core. Now the cut-off is applied to VdW with soft-core.
Fixes possible NaN in free energy kernel with r close to 0.
Replaced a confusing grompp warning with PME and soft-core by a note.
Removed an invalid grompp warning about soft-core and twin-range.
Fixes #1146

Change-Id: I79a06b20158df2bce575808b6e31e660163bd307

11 years agoMerge release-4-5-patches into release-4-6
Mark Abraham [Tue, 19 Feb 2013 13:01:57 +0000 (14:01 +0100)]
Merge release-4-5-patches into release-4-6

Avoided the 4.5.6 release patch by using
git merge -s ours 10b109beb3ffa95e37386d38140d2c0fd7769f20
git merge release-4-5-patches
tree=$(git log -1 HEAD --pretty=%T)
git reset --hard $(git cat-file commit HEAD | sed '1,/^$/d' | \
    git commit-tree $tree -p 2b05689 -p 8d6cc14)
Conflict resolution was straightforward, except that the vsite
code was somewhat reorganized in the meantime.

Conflicts:
        Makefile.am
        src/contrib/Makefile.am
        src/contrib/addquote.c
        src/gmxlib/copyrite.c
        src/mdlib/vsite.c

Change-Id: Ib90ac4e55fac308c433b15c344caf5ee41f0146c

11 years agofixed GPU particle gridding performance issue
Berk Hess [Thu, 14 Feb 2013 21:27:17 +0000 (22:27 +0100)]
fixed GPU particle gridding performance issue

The scaling factor for the grid binning for the GPU pair search
was set incorrectly, which made the binning 50% slower.

Change-Id: I146592c37094a3d81a7ae50b3903fcc615e748d5

11 years agoMade g_tune_pme work with 4.6 if user sets "-p" command line option
Carsten Kutzner [Fri, 15 Feb 2013 10:51:09 +0000 (11:51 +0100)]
Made g_tune_pme work with 4.6 if user sets "-p" command line option

Added "-p" (performance statistics log file) to the list of files that
must not appear on the mdrun command line at launch time. (This would lead
mdrun to refuse to launch due to an unknown command line option.)

Change-Id: I268bbf6513709866b9a6a8d81c97706ccc832988

11 years agoFix CMake namespace pollution
Mark Abraham [Sat, 9 Feb 2013 16:35:11 +0000 (17:35 +0100)]
Fix CMake namespace pollution

When gmxManageMPI finds MPI and CMake version is less than 2.8.10,
the variable file_cmd ends up in the non-advanced CMake namespace.
Made it advanced in those cases.

Change-Id: I79fd46e1b66d6c5ab008973e654092939eeae4dc

11 years agoIssue errors/warnings for ICC before 12.0.0
Roland Schulz [Mon, 21 Jan 2013 22:58:46 +0000 (17:58 -0500)]
Issue errors/warnings for ICC before 12.0.0

ICC 11.1 with SSE4.1 gives incorrect results, so CMake will
refuse to build with that minor version. Given that we
currently don't know anything specific and 11 is old, we
warn with any version older than 12.0.0. If more information
is available later, this warning can be made more specific.

Fixes #1126

Change-Id: I3c37844c69ce34769187f288c43cd8aef4086c53

11 years agobugfix for md-vv + nose-hoover + (nstcalcenergy > nsttcouple)
Michael Shirts [Thu, 14 Feb 2013 03:01:14 +0000 (22:01 -0500)]
bugfix for md-vv + nose-hoover + (nstcalcenergy > nsttcouple)

Need global communication the step befre nsttcouple
in order to get the kinetic energy right, because
of the offsets velocity verlet from leapfrog.

Fixes redmine #1129

Change-Id: Ia6157db8304cc4fb8cccbda0ac3b949e800efeb5

11 years agoUse explicit kernel pointer typecasts
Mark Abraham [Tue, 29 Jan 2013 13:29:28 +0000 (14:29 +0100)]
Use explicit kernel pointer typecasts

Some compilers don't like implicit casts from function pointers
to void*, so if we're going to be dirty, dirty people, we may
as well come clean about it.

Change-Id: Ia071c6a10a4ec20b83ddb13e4f0cd2a7b139264f

11 years agoMerge "Minor fixes to Verlet kernels" into release-4-6
Mark Abraham [Fri, 15 Feb 2013 22:32:16 +0000 (23:32 +0100)]
Merge "Minor fixes to Verlet kernels" into release-4-6

11 years agoBump shared object version to 7
Mark Abraham [Fri, 15 Feb 2013 22:23:20 +0000 (23:23 +0100)]
Bump shared object version to 7

Almost certainly the ABI is binary incompatible for the 4.5.5 -> 4.5.6
transition, but we didn't think of handling the issue at the time.
So we can do so now.

Refs #1147

Change-Id: I15f978547d7fbcc9a72454067825c71c0ffb459a

11 years agoMerge "improved PME load balancing reporting" into release-4-6
Mark Abraham [Fri, 15 Feb 2013 18:29:33 +0000 (19:29 +0100)]
Merge "improved PME load balancing reporting" into release-4-6

11 years agoMerge "Re-generated group kernels" into release-4-6
Mark Abraham [Fri, 15 Feb 2013 16:51:47 +0000 (17:51 +0100)]
Merge "Re-generated group kernels" into release-4-6

11 years agoMerge "Unwinds flag dependencies in g_order." into release-4-6
Mark Abraham [Fri, 15 Feb 2013 14:54:21 +0000 (15:54 +0100)]
Merge "Unwinds flag dependencies in g_order." into release-4-6

11 years agoMerge "Fixed an initialization issue in gmx_dipoles." into release-4-6
David van der Spoel [Fri, 15 Feb 2013 12:41:47 +0000 (13:41 +0100)]
Merge "Fixed an initialization issue in gmx_dipoles." into release-4-6

11 years agoMerge "use legacy CUDA non-bonded kernels only on Fermi" into release-4-6
Mark Abraham [Fri, 15 Feb 2013 12:38:24 +0000 (13:38 +0100)]
Merge "use legacy CUDA non-bonded kernels only on Fermi" into release-4-6

11 years agoMerge "More descriptive Verlet kernel macro parameter names" into release-4-6
Mark Abraham [Fri, 15 Feb 2013 12:38:05 +0000 (13:38 +0100)]
Merge "More descriptive Verlet kernel macro parameter names" into release-4-6

11 years agoFixed an initialization issue in gmx_dipoles.
Sander Pronk [Fri, 15 Feb 2013 12:12:53 +0000 (13:12 +0100)]
Fixed an initialization issue in gmx_dipoles.

This fixes a warning in gcc 4.7

Change-Id: Ief8a057dff24c9a61c20301215a7009c2a16e74a

11 years agoimproved PME load balancing reporting
Berk Hess [Tue, 5 Feb 2013 11:55:52 +0000 (12:55 +0100)]
improved PME load balancing reporting

With the group cut-off scheme, rlist was reported incorrectly.
Added a note in log and stderr when PME load balancing increases
the non-bonded workload by more than 50%.

Change-Id: I7d46f5a110b24cbb40b6b9cb53a1d90715c7c5ec

11 years agoMinor fixes to Verlet kernels
Mark Abraham [Mon, 28 Jan 2013 15:48:08 +0000 (16:48 +0100)]
Minor fixes to Verlet kernels

* clarified some comments
* documented use of nbfp_s4 better

Change-Id: I0848f2e570daef881368a45d4a429a3f80fde81b

11 years agoUnwinds flag dependencies in g_order.
Peter Kasson [Tue, 22 Jan 2013 20:18:43 +0000 (12:18 -0800)]
Unwinds flag dependencies in g_order.

A feature added a while back to g_order had a flag dependency
(-calcdist forced -radial). This patch attempts to unwind that
dependency and allow -calcdist independent of -radial.

Fixed help message for -calcdist

Change-Id: I8c368f9109ee305d79a3c0c1b6838439598a4e89

11 years agoMerge "Fix and add Verlet kernel error messages" into release-4-6
Mark Abraham [Fri, 15 Feb 2013 08:41:16 +0000 (09:41 +0100)]
Merge "Fix and add Verlet kernel error messages" into release-4-6

11 years agoMerge "Fix code bug with thread-MPI on PowerPC with xlc" into release-4-6
Sander Pronk [Wed, 13 Feb 2013 21:24:25 +0000 (22:24 +0100)]
Merge "Fix code bug with thread-MPI on PowerPC with xlc" into release-4-6

11 years agoMerge "Fixed the MPI recognition string in g_tune_pme" into release-4-6
Roland Schulz [Wed, 13 Feb 2013 21:24:19 +0000 (22:24 +0100)]
Merge "Fixed the MPI recognition string in g_tune_pme" into release-4-6

11 years agoMerge "improve setup of multiple tf tables" into release-4-6
Roland Schulz [Wed, 13 Feb 2013 21:22:49 +0000 (22:22 +0100)]
Merge "improve setup of multiple tf tables" into release-4-6

11 years agoPlain text quotes.
David van der Spoel [Sat, 15 Dec 2012 10:11:43 +0000 (11:11 +0100)]
Plain text quotes.

And removed 5 gurgles and 2 bromacses - replaced the bromacses
by new and exciting ones.
Change-Id: Ibdad70d6ceb0caa52be351762704a677e9dddba8

11 years agoFix code bug with thread-MPI on PowerPC with xlc
Mark Abraham [Tue, 29 Jan 2013 12:37:30 +0000 (13:37 +0100)]
Fix code bug with thread-MPI on PowerPC with xlc

Change-Id: I8254ae915fdd24e02cf8f07789c69c3d4b35525e

11 years agoMore descriptive Verlet kernel macro parameter names
Mark Abraham [Sun, 3 Feb 2013 10:40:06 +0000 (11:40 +0100)]
More descriptive Verlet kernel macro parameter names

Also more consistent with SIMD kernel usage

Change-Id: Ia8dd723075b5c289d0171e3d34ecd51a0fc152f4

11 years agoFix and add Verlet kernel error messages
Mark Abraham [Sat, 2 Feb 2013 11:53:19 +0000 (12:53 +0100)]
Fix and add Verlet kernel error messages

Handles default cases better

Change-Id: I45d8ebd73184630180e6ed87d54c3a8fc3cbef73

11 years agoRe-generated group kernels
Mark Abraham [Wed, 13 Feb 2013 15:07:33 +0000 (16:07 +0100)]
Re-generated group kernels

This should be a null operation, but I had to fix copyright
years in python scripts and it had the following effects:
* removed some debugging printf statements in GB C kernels
* minor whitespace change now consistent between template and output

Change-Id: I3c3fe9be625ca4f7ade6871fc105fd80899e5a70

11 years agoAdded extra check in for setaffinity support.
Sander Pronk [Wed, 13 Feb 2013 14:24:22 +0000 (15:24 +0100)]
Added extra check in for setaffinity support.

The Linux version of tMPI_Thread_setaffinity_support() did not check
whether pthread_setaffinity_np() would actually work on the system
it's running on. It now checks by running pthread_getaffinity_np()
and checking its return value.

This is relevant for, for example, Bluegene systems which don't
support explicit affinity setting.

Change-Id: I4833e7384067f897013d6fa4d0b35963ae740fcf

11 years agoFixed the MPI recognition string in g_tune_pme
Carsten Kutzner [Wed, 13 Feb 2013 11:41:46 +0000 (12:41 +0100)]
Fixed the MPI recognition string in g_tune_pme

Since the identifying string "NNODES="is not printed any more by MPI-enabled
mdrun, we now use the "MPI library:" string instead.

Change-Id: Ibbcc0678a7c89057743129aa664fad8bb68f1941

11 years agoMerge "Updated FindGSL.cmake to version written by me." into release-4-6
Roland Schulz [Wed, 13 Feb 2013 01:15:23 +0000 (02:15 +0100)]
Merge "Updated FindGSL.cmake to version written by me." into release-4-6

11 years agoMerge "PME-only counters are now reset with -maxh -resethway" into release-4-6
Mark Abraham [Tue, 12 Feb 2013 19:58:14 +0000 (20:58 +0100)]
Merge "PME-only counters are now reset with -maxh -resethway" into release-4-6

11 years agoUpdated FindGSL.cmake to version written by me.
Christoph Junghans [Wed, 16 Jan 2013 21:25:52 +0000 (14:25 -0700)]
Updated FindGSL.cmake to version written by me.

relates to #815

Change-Id: I323f113b4016ef232124fcdcc985d0526bc8a880

11 years agoMerge "fixed bug with sd integrator and OpenMP threading" into release-4-6
Mark Abraham [Tue, 12 Feb 2013 19:56:14 +0000 (20:56 +0100)]
Merge "fixed bug with sd integrator and OpenMP threading" into release-4-6

11 years agoMerge "Fix CUDA_HOST_COMPILER flag name" into release-4-6
Mark Abraham [Tue, 12 Feb 2013 19:43:02 +0000 (20:43 +0100)]
Merge "Fix CUDA_HOST_COMPILER flag name" into release-4-6

11 years agouse legacy CUDA non-bonded kernels only on Fermi
Szilárd Páll [Wed, 16 Jan 2013 20:05:28 +0000 (21:05 +0100)]
use legacy CUDA non-bonded kernels only on Fermi

This solves a remainig todo as well as clarifies some variable naming
and documentation.

Change-Id: Ifad80cb7e740d1329a9c84df992d899400efeca2

11 years agoMerge "Typo fix" into release-4-6
Roland Schulz [Sat, 9 Feb 2013 18:18:10 +0000 (19:18 +0100)]
Merge "Typo fix" into release-4-6

11 years agoTypo fix
Mark Abraham [Sat, 9 Feb 2013 17:20:36 +0000 (18:20 +0100)]
Typo fix

Change-Id: Id5bb7297adbc38909aa184b6d6a222aa719f90be

11 years agoFix CUDA_HOST_COMPILER flag name
Roland Schulz [Fri, 8 Feb 2013 20:09:06 +0000 (15:09 -0500)]
Fix CUDA_HOST_COMPILER flag name

Fixes #1143

Change-Id: If34f27aa8fa73420ac85d4ec6997e8ebe78261d4

11 years agoMerge "added back cuda.h includes to fix missing CUDA_VERSION" into release-4-6
Mark Abraham [Fri, 8 Feb 2013 15:16:30 +0000 (16:16 +0100)]
Merge "added back cuda.h includes to fix missing CUDA_VERSION" into release-4-6

11 years agoMerge "Fixed flop counting for generic cg/adress kernels" into release-4-6
Mark Abraham [Fri, 8 Feb 2013 15:16:17 +0000 (16:16 +0100)]
Merge "Fixed flop counting for generic cg/adress kernels" into release-4-6

11 years agoMerge "cmake: fixed typo in template CMakeLists.txt" into release-4-6
Mark Abraham [Fri, 8 Feb 2013 15:16:05 +0000 (16:16 +0100)]
Merge "cmake: fixed typo in template CMakeLists.txt" into release-4-6

11 years agofixed bug with sd integrator and OpenMP threading
Berk Hess [Wed, 6 Feb 2013 15:21:24 +0000 (16:21 +0100)]
fixed bug with sd integrator and OpenMP threading

This could cause small error in the integration or a segv.
This bug was introduced with d120c370
Fixes #1138

Change-Id: I4ae2c0c857cb68e796d01c5b2e0ea42beaf13d4a

11 years agoMerge "Documented what GMX_CPU_ACCELERATION really means" into release-4-6
Christoph Junghans [Tue, 5 Feb 2013 23:52:49 +0000 (00:52 +0100)]
Merge "Documented what GMX_CPU_ACCELERATION really means" into release-4-6

11 years agoimprove setup of multiple tf tables
Christoph Junghans [Wed, 23 Jan 2013 20:57:13 +0000 (13:57 -0700)]
improve setup of multiple tf tables

* buffer string was not initialized
* put tf table info in log instead of on the screen

Change-Id: I11ecdcd8c9a18a8931dbccef528ff02a74959ee1

11 years agoMerge "improved load balancing on the GPU" into release-4-6
Mark Abraham [Tue, 5 Feb 2013 19:27:04 +0000 (20:27 +0100)]
Merge "improved load balancing on the GPU" into release-4-6

11 years agoDocumented what GMX_CPU_ACCELERATION really means
Mark Abraham [Tue, 5 Feb 2013 19:00:46 +0000 (20:00 +0100)]
Documented what GMX_CPU_ACCELERATION really means

Discussion arose in I0848f2e570daef881368a45d4a429a3f80fde81b

Change-Id: Ib99ab86ff7d8acf913121decb00756f47327f046

11 years agoimproved load balancing on the GPU
Berk Hess [Fri, 1 Feb 2013 09:22:09 +0000 (10:22 +0100)]
improved load balancing on the GPU

For the GPU, small pair list entries are now sorted to the end.
The improves performance by 5 to 20%.

Change-Id: I25e5efeb813ad5dde48f0955366519db699f21a2

11 years agofixed segv in init_forcerec with fr->hwinfo=NULL
Berk Hess [Mon, 4 Feb 2013 12:44:31 +0000 (13:44 +0100)]
fixed segv in init_forcerec with fr->hwinfo=NULL

This caused genion and g_disre to segv with the Verlet scheme.
Fixes #1136

Change-Id: I9059b0e108c80ea86567a568e9f488b4db500eee

11 years agoFixed flop counting for generic cg/adress kernels
Christoph Junghans [Sat, 2 Feb 2013 03:02:04 +0000 (20:02 -0700)]
Fixed flop counting for generic cg/adress kernels

Change-Id: I95f2a68a0de64f479820143fdba09ea09805fa93

11 years agocmake: fixed typo in template CMakeLists.txt
Christoph Junghans [Sat, 2 Feb 2013 01:46:46 +0000 (18:46 -0700)]
cmake: fixed typo in template CMakeLists.txt

* prevented build after make clean, when build was used before

Change-Id: I0ba88b36752801c3d492d46c71327fb966edb508

11 years agoPME-only counters are now reset with -maxh -resethway
Berk Hess [Wed, 30 Jan 2013 09:57:51 +0000 (10:57 +0100)]
PME-only counters are now reset with -maxh -resethway

Fixes #1125
For clarity gmx_pme_recv_q_x now returns and enum
and gmx_pme_send_switch is renamed to gmx_pme_send_switchgrid.

Change-Id: I21952780f0b719378d947a79324c844ccde7160a

11 years agoAdResS: fixes bug in the neighbour list routine
Sebastian Fritsch [Mon, 28 Jan 2013 19:49:40 +0000 (20:49 +0100)]
AdResS: fixes bug in the neighbour list routine

* coarse-grained interactions were not excluded in the all-atom region
* Some atomistic interactions were not excluded in the coarse-grained region

Change-Id: Ib4ac145722367cf139ad3d23b86f0de1ca9d6caf

11 years agofixed copyright of adress kernel
Christoph Junghans [Fri, 18 Jan 2013 21:46:05 +0000 (14:46 -0700)]
fixed copyright of adress kernel

Change-Id: I21816431b4f77210a93dedc7a265c10f074502c4

11 years agoMerge "Moved a few comments back into position" into release-4-6
Christoph Junghans [Fri, 25 Jan 2013 21:06:55 +0000 (22:06 +0100)]
Merge "Moved a few comments back into position" into release-4-6

11 years agofix invalid read in construct_vsites
Han Wang [Wed, 23 Jan 2013 21:02:22 +0000 (14:02 -0700)]
fix invalid read in construct_vsites

* aj is not available in the F_VSITEN case hence cannot be read in
  the common part of the function

Change-Id: Ic98dfb91eb095aea933abc3dbfb4f5cc10bf4166

11 years agoadded back cuda.h includes to fix missing CUDA_VERSION
Szilard Pall [Thu, 24 Jan 2013 01:27:09 +0000 (02:27 +0100)]
added back cuda.h includes to fix missing CUDA_VERSION

nvcc doesn't define its version identity macro, CUDA_VERSION, by default,
but this needs to be pulled in through cuda.h. Quite weird and
non-standard behavior, but as swithing to legacy kernels and certain
kernel optimizations are broken now, to fix these we need to include
cuda.h.

Change-Id: I4836335e97a287e18e595c160819db2e46a8d798

11 years agoMerge "Illegal acceleration defaulted to QPX" into release-4-6
Mark Abraham [Tue, 22 Jan 2013 18:17:52 +0000 (19:17 +0100)]
Merge "Illegal acceleration defaulted to QPX" into release-4-6

11 years agoMerge "Re-licenced FindGROMACS.cmake (was written by me)" into release-4-6
Mark Abraham [Tue, 22 Jan 2013 16:41:00 +0000 (17:41 +0100)]
Merge "Re-licenced FindGROMACS.cmake (was written by me)" into release-4-6

11 years agoIllegal acceleration defaulted to QPX
Roland Schulz [Mon, 21 Jan 2013 21:34:35 +0000 (16:34 -0500)]
Illegal acceleration defaulted to QPX

Change-Id: I1bfd9c7bc7bac68944fd912bda84e8bd03af1fbc

11 years agoMerge "Fix TOUPPER usage" into release-4-6
Mark Abraham [Tue, 22 Jan 2013 16:40:12 +0000 (17:40 +0100)]
Merge "Fix TOUPPER usage" into release-4-6

11 years agoMerge "Updated html manual to 4.6" into release-4-6
Mark Abraham [Tue, 22 Jan 2013 16:39:55 +0000 (17:39 +0100)]
Merge "Updated html manual to 4.6" into release-4-6

11 years agoMerge "added missing cleanup of some nbnxn_cuda data" into release-4-6
Mark Abraham [Tue, 22 Jan 2013 15:31:01 +0000 (16:31 +0100)]
Merge "added missing cleanup of some nbnxn_cuda data" into release-4-6