Szilárd Páll [Mon, 29 Jul 2019 16:24:46 +0000 (18:24 +0200)]
Store wcycle pointer in nonbonded_verlet_t
Removed wcycle arguments from method calls that no longer need it and
moved a few instances of cycle counting into the nbmxn module.
Change-Id: Ic5646b3bb85ed2c66137e9db7bd70822df95042b
Szilárd Páll [Wed, 10 Jul 2019 18:48:16 +0000 (20:48 +0200)]
Fixes for F buffer ops change
Change-Id: I8880ff91156c983560ee2173748faa8a4b189817
Berk Hess [Tue, 30 Jul 2019 11:50:13 +0000 (13:50 +0200)]
Remove need for GROMOS FF in test
The methanol + SPC test system in the simulation database now
incorpates the needed GROMOS force field parameters in the top file
to avoid dependence on the deprecated force field files.
Change-Id: Ib039c09fd38dfc153aed0787ecf14991ccb795cf
Pascal Merz [Mon, 27 May 2019 19:36:35 +0000 (13:36 -0600)]
Rename Simulator to LegacySimulator
In view of the new modular simulator, we will want an interface for
simulators rather than a single rigid implementation. As we will keep
the current implementation around for a while, we're renaming it here
to free up the generic "Simulator" term.
Change-Id: Iaa5cca332130b8324e11206e2044af799a52f9c0
Szilárd Páll [Fri, 26 Jul 2019 18:31:49 +0000 (20:31 +0200)]
Move cycle counting into the GPU bonded module
Additionally:
- removed some outdated TODOs;
- renamed accumulateEnergyTerms() to waitAccumulateEnergyTerms()
to reflect that it does a blocking wait for results.
Change-Id: I1bd7f1ddecea94ff2696721c9f2d6eac45bea738
Szilárd Páll [Mon, 15 Jul 2019 17:38:58 +0000 (19:38 +0200)]
Report when development feature flags are used
On order to help identify runs that use development features triggered
using environment variables, this change adds reporting of these to both
log and standard output.
Change-Id: I6d554af25e3b17b38f3826efbf2a5179bd6d3b89
Szilárd Páll [Fri, 26 Jul 2019 17:35:56 +0000 (19:35 +0200)]
Move end of step GPU task launches to a function
Code movement mostly to improve readability; also moved the misplaced
call to pme_gpu_reinit_computation() which should be called after the
GPU force buffer reduction.
Change-Id: I4dc602b16b67a3f79ee3b8802a6f5d0d39a8f33b
Berk Hess [Wed, 26 Jun 2019 16:51:24 +0000 (18:51 +0200)]
Use invsqrt and cbrt in free energy kernel
This speeds up the soft-core free-energy kernel by 15%.
Change-Id: Id97226c8611d56312d89df94dec29eeb766a45a8
Paul Bauer [Thu, 30 Aug 2018 13:35:06 +0000 (15:35 +0200)]
Add wrapper for outputadapter methods
Adds a wrapper for containing all the commandline options needed to
initialize the different outputadapter methods.
Change-Id: I5ea750203f3e493394ec7e62f385b05b20e442a7
Berk Hess [Fri, 26 Jul 2019 09:35:31 +0000 (11:35 +0200)]
Remove LJ Ewald correction tables
The LJ Ewald correction tables have been removed. This has
the (positive) side effect that the table spacing for Coulomb
Ewald is no longer dependent on the LJ Ewald settings.
Also removed rtab in several places, since it was only used
with the group cutoff scheme.
Change-Id: Ib7d75071bba24508adb94c5d50ac3629d63db10c
Berk Hess [Fri, 26 Jul 2019 13:19:32 +0000 (15:19 +0200)]
Add initialization to t_nrnb
Also fixed a memory leak.
Change-Id: I9c4bd3ed1919c82cf26d1181ea98d4c2dcc37b0b
Berk Hess [Fri, 26 Jul 2019 11:29:35 +0000 (13:29 +0200)]
Removed generalized reaction-field
This only worked correctly with the group scheme and charge groups.
Although this could be implemented by using net charges of molecules,
having the interaction function depend on the system volume and
temperature complicates the code and correctness checking.
Change-Id: If29b28772cd83524c5984f64cfcd88cc4060ca36
Berk Hess [Fri, 26 Jul 2019 06:52:09 +0000 (08:52 +0200)]
Convert Ewald correction tables to C++
This change introduces a separate struct for the non-bonded
Ewald correction tables that uses std::vector instead of
manual memory management.
Although there are new (unused) table classes, the reorganization
is needed now to convert interaction_const_t to proper C++.
This change is only refactoring.
Change-Id: I9dbdcd401c8a3d9674775a25117663cbe86ecfb3
Berk Hess [Thu, 25 Jul 2019 13:44:17 +0000 (15:44 +0200)]
Add default initialization to interaction_const_t
Also changed the manual destruction to a destructor.
Change-Id: I4c075c7f17b3376edd3b60bc25108f57cc404c88
Christian Blau [Thu, 18 Jul 2019 09:08:11 +0000 (11:08 +0200)]
Fix failing check-source target on python3.6
When the locale is not set to UTF-8 encoding, check-source fails.
This fixes #3039
Change-Id: I79b222ae9f839123f3e7411c5d73249462aafab0
Berk Hess [Wed, 24 Jul 2019 13:56:24 +0000 (15:56 +0200)]
Remove sh_invrc6 from interaction_const_t
Change-Id: Ib903b9b0dc3a85f290e084db87f102b9c05c2f56
Artem Zhmurov [Mon, 27 May 2019 16:56:39 +0000 (18:56 +0200)]
Remove PImpl scaffolding from CUDA version of SETTLE
GPU version of SETTLE was implemented as a class with private
implementation so it will be possible to initialize on
non-CUDA hosts. Now, the implementation can be hidden
inside the Update and Constraints PImpl so that the CUDA
specific types and calls can be exposed in SETTLE and
private implementation is no longer needed there.
Refs #2816, #2888
Change-Id: I4c78f2629be34b42bb5f4f7d34970c3e41515691
Berk Hess [Mon, 22 Jul 2019 09:24:35 +0000 (11:24 +0200)]
Split canDetectGpus()
The function canDetectGpus() was used in a higher and lower level
fashion. The original canDetectGpus() is renamed to
isGpuDetectionFunctional(). Added canPerformGpuDetection() which also
checks for a GPU build and the env.var. GMX_DISABLE_GPU_DETECTION.
This also fixes an inconsistency error in the PME GPU unit tests
when running with GMX_DISABLE_GPU_DETECTION.
Change-Id: If762b2a108eedeb1bda0a2af9afa8a418aa93e14
Szilárd Páll [Mon, 15 Jul 2019 15:46:20 +0000 (17:46 +0200)]
Only stop the CUDA profiler if there are devices
If there are no CUDA devices the unconditionally called cudaProfilerStop
returns with an error. This change makes the the stop API call
conditional to avoid the corner-case assertion error in a profiler run
with no GPUs.
Change-Id: Ife306c927d163163511a2ebf06a7efe6d4122490
Kevin Boyd [Mon, 22 Jul 2019 19:34:17 +0000 (15:34 -0400)]
Add transpose operation to Matrix3x3
refs #2976
Change-Id: I8f0ce04b18be1e8b119de02cb6b52b3314cd67eb
Paul Bauer [Thu, 4 Jul 2019 09:00:43 +0000 (11:00 +0200)]
Fix doxygen in coordianteio files
Had the wrong order for the doxygen statements and corrected that.
Change-Id: Iead4ca5e6b9747f2dcb435f610f8e3dc042da5aa
Paul Bauer [Thu, 30 Aug 2018 13:31:08 +0000 (15:31 +0200)]
Add outputadapters (3/3)
Add outputadapters based on IOutputAdapter that can be used to set the
output precision and change the time information included in a
t_trxframe.
Change-Id: I3b0b5e975235e4b3e93a159e1a60490ba9d4386b
Christian Blau [Tue, 23 Jul 2019 15:26:38 +0000 (17:26 +0200)]
Fix HTML source documenation error caused by GaussTransform3D
In doxygen/html-full/annotated.xhtml, doxygen generated an invalid
xhtml file with the following error on line 724 at column 493:
Opening and ending tag mismatch: td line 0 and p
This patch fixes the error by changing the math rendering from
multi-line to single line.
It further removes superfluous * in the verbatim comment of same
class.
Change-Id: I80996743743758095ceac78c519fbec88da9f772
M. Eric Irrgang [Tue, 16 Jul 2019 14:42:51 +0000 (17:42 +0300)]
Remove one of the examples from sample_restraint.
Remove the old simple harmonic spring potential. It is not useful and
presents a maintenance headache as we modify the plugin interfaces.
Take this opportunity to clarify or simplify some naming.
Refs #3027
Change-Id: Ic0c59b0f2b9391beac48f64cabba8dbf05f0cd63
Mark Abraham [Wed, 24 Jul 2019 11:26:19 +0000 (13:26 +0200)]
Remove tests of make_unique
These made sense when we had compat::make_unique, but we should
not be maintaining code for testing std components.
Change-Id: I794e33638ed97331453bb28da2fbb7b944e72ed2
M. Eric Irrgang [Wed, 10 Jul 2019 15:55:24 +0000 (18:55 +0300)]
Begin new sample_restraint development.
* Update the sample MD plugin project for its new home in the GROMACS
source tree and the gmxapi Python package that is now part of the
GROMACS distribution (as opposed to gmxapi 0.0.7 hosted at
https://github.com/kassonlab/sample_restraint)
* Update googletest infrastructure to use CMake 3.9 utilities.
* Download googletest and pybind sources during docker CI build.
* Download binary test data from GitHub.
Future updates are necessary both for using the subtree in Jenkins
builds (to avoid downloading or using redundant copies of dependencies)
and for using the subtree as a starting point for an external project
(in which case the external dependencies _should_ be bundled).
Also, requires Change-Id: I44a66646eceb191a2ec755169671af326f22526a to
pass all CI tests.
Refs #3027
Change-Id: I4dd699168d6bc8e84e7de4d3b5b7b1cd6bd21090
M. Eric Irrgang [Wed, 24 Jul 2019 09:27:44 +0000 (12:27 +0300)]
Fix gmxapi notebook image and sample_restraint docker build stage.
Resolve conflicting logic from upstream merges.
Change-Id: I2fb42e602ce73d5849cf435f713f2c35101674e2
M. Eric Irrgang [Tue, 23 Jul 2019 19:44:56 +0000 (22:44 +0300)]
Clarify use of Python venv in gmxapi docker images.
Refs #2756
Change-Id: I182c83be424fcf134bc0f3ce7c41a58c6fb1f07e
M. Eric Irrgang [Wed, 26 Jun 2019 11:35:23 +0000 (14:35 +0300)]
Improve gmxapi docker testing environment.
* Generalize venv directory.
* Add some usage notes.
* Better support alternative users in executing containers.
* Remove stale TODO.
* Allow for passing additional arguments to pytest.
Refs #2756
Change-Id: I9f67d46aa93fb789825cb78cc2a95b24f7dfccaa
M. Eric Irrgang [Mon, 22 Jul 2019 11:30:02 +0000 (14:30 +0300)]
Improve docs for gmxapi Python environment preparation.
Try to account for differences in system environments with additional
notes on accessing different versions of Python interpreters and on
bootstrapping pip.
Change-Id: I7acea811d86e801b1c9989b09041627c9be718c9
Mark Abraham [Tue, 23 Jul 2019 08:44:35 +0000 (10:44 +0200)]
Update developer guide
Noted best practice for using forward declarations in headers.
Change-Id: Ib6411c43d93481a45d4f065b8dd6c8141f199b61
M. Eric Irrgang [Wed, 17 Jul 2019 15:21:22 +0000 (18:21 +0300)]
Don't install sample restraint package by default.
When building the sample_restraint project as part of a GROMACS build,
do not installed the myplugin Python package as part of the `install`
target. This is almost undoubtedly not what the user intended.
However, the package is still built when GMX_PYTHON_PACKAGE=ON to help
catch errors and facilitate testing by developers or automated systems
(i.e. Jenkins).
Refs #3027
Change-Id: I934fcc9f8f6f3f749435c5a85a904006e48d1d81
Kevin Boyd [Fri, 12 Jul 2019 15:50:02 +0000 (11:50 -0400)]
Modernize syntax of enable_if and traits to use _t helpers
From c++ 14, "typename std::enable_if<something>::type"
can be reduced to
"std::enable_if_t<something>"
as can traits with ::type
Note that when we switch to C++ 17, a number of traits with
::value can be similarly shorthanded with _v
Change-Id: I92d4502c51ac613fa81f130ed3127c812c4e0a6f
Szilárd Páll [Mon, 15 Jul 2019 11:26:36 +0000 (13:26 +0200)]
Conditionally pin GPU-related grid data
Data that is transferred to the GPU when the buffer ops is offloaded is
now only pinned when the nonbonded module uses GPU offload avoidign the
runtime errors encountered when a GPU-enabled build does not detect a
GPU and therefore the CUDA runtime refuses to register the memory.
Refs #2817 #2934
Change-Id: Iabbc0d9f37fad0e88cd39a078af1346e8f713ec1
Alan Gray [Fri, 8 Mar 2019 09:05:47 +0000 (01:05 -0800)]
F buffer operations in CUDA
This patch performs GPU buffer ops for force buffers.
Enable with GMX_USE_GPU_BUFFER_OPS env variable.
Currently, the H2D transfer of the force buffer is switched on with
haveSpecialForces || haveCpuBondedWork || haveCpuPmeWork,
where haveCpuPmeWork is true even when useGpuPme == true
until on-GPU PME-nonbonded reduction is added in follow-up.
TODO: enable PME reduction in GPU buffer ops and remove associated H2D
transfer
Implements part of #2817
Change-Id: Ice984425301d24bac1340e883698244489cd686e
M. Eric Irrgang [Tue, 16 Jul 2019 15:09:07 +0000 (18:09 +0300)]
Improve some documentation formatting and spelling.
Change-Id: Ic59eb3407fda2dbd341fb4ce6235d7c5c99c7630
M. Eric Irrgang [Thu, 11 Jul 2019 16:59:53 +0000 (19:59 +0300)]
Remove extraneous file.
A file was overlooked in Change-Id: I0d847bc2490935e0ddc3170b674d7460a5a08506
Refs #3033
Change-Id: I4fb6af0b6d51b477091c6305872c0c7124e754d6
Paul Bauer [Thu, 27 Jun 2019 15:13:39 +0000 (17:13 +0200)]
Extend MDLogger levels
Added logging levels for errors and debug information.
Extended tests to check those.
Refs #2999
Change-Id: I8b23de7baf3555ffc89896e68c1998608b7b71cf
Paul Bauer [Wed, 10 Jul 2019 14:07:20 +0000 (16:07 +0200)]
Move methods out from trjconv
Moves some functions out of trjconv to allow access in other tools as
well. This change is just code movement.
Change-Id: I37e9f8d87862e6ad6666704dc85d74b09b144f49
Mark Abraham [Wed, 10 Jul 2019 18:33:05 +0000 (20:33 +0200)]
Add gpubufferops test matrix
This commit teaches Jenkins to understand how to enable the GPU buffer
ops environment variable that is the feature flag for a developmental
feature. The new capability is used in a releng build configuration
matrix that releng can be taught to trigger on demand with
[JENKINS] gpubufferops
comments in gerrit
Change-Id: I30ca4a5d84fe136a657be7a134829f4c696c5d85
Szilárd Páll [Wed, 24 Apr 2019 12:03:39 +0000 (14:03 +0200)]
Use HostVector for Grid/GridSet data need on-GPU
Grid.cxy_na_, Grid.cxy_ind_, GridSet.cells and GridSet.atomIndices
have been converted from std::vector to gmx::HostVector. This allow
the code to pin the HostVector when X buffer ops is used and to
eliminate the hacky pin/unpin in CUDA buffer ops functions.
Part of #2934
Refs #2817
Change-Id: Icca21dd076128ec582f805ed96e253dfab461270
Mark Abraham [Tue, 2 Jul 2019 07:46:13 +0000 (09:46 +0200)]
Fix pre-submit build matrix and description
Pre-submit should not fail because the ftp server is not reachable, so
moved it to nightly matrix. Also recorded the intent to test
such a build.
Also fixed that the intent to test AVX_128_FMA in mixed precision was
broken in Ib2c5e14e5d89665130ae8d6c749c3230ccfe72a9. There's no
ability to test this SIMD in pre-submit without wider changes, and their
is neither large risk nor consequence of breaking it, so nightly is OK.
Refs #2809, #2629
Change-Id: I9cd41a359ba662a8a9529540c50c9a7ef546cd29
Kevin Boyd [Wed, 10 Jul 2019 22:01:59 +0000 (18:01 -0400)]
Use const views in matrix determinant calculation
Reduces scope of the matrix.h implementation to just reals
Change-Id: Ide66967b35ffc58b3e61c403d3ce1862b5f5b8ca
Kevin Boyd [Tue, 2 Jul 2019 03:31:02 +0000 (23:31 -0400)]
Add trace functionality to 3x3 matrices
Replacement for vec.h trace
Change-Id: I5c452cb36bab277ffcb3f2f9645d3d025a1805a8
Pascal Merz [Mon, 1 Jul 2019 05:45:18 +0000 (23:45 -0600)]
Get rid of clang-tidy warning in do_md/rerun/mimic
This change silences most of the clang-tidy warnings
"Clang-Tidy: Use of a signed integer operand with a binary
bitwise operator" in do_md, do_rerun and do_mimic. This warning
is triggered when using a signed integer with a binary bitwise
operator, as we do all over the code when encoding / decoding
flags. Such warnings are popping up in other parts of the code,
but this change takes only care of the ones which can be
silenced by changes in the simulator functions and the bit shift
definitions.
Change-Id: I7345e987e91051c9688cb2e8a96026682f025b6b
Pascal Merz [Mon, 1 Jul 2019 00:10:42 +0000 (18:10 -0600)]
Remove dependence of compute_globals on t_state
This change makes the different function related to compute_globals
request the fields of t_state explicitly, instead of simply taking
a pointer to the full t_state object. This simplifies refactoring
the t_state object in future commits. This change also clarifies why
both state->box and an additional parameter `matrix box` is required -
the latter one is the box at the last complete timestep in do_md and
has been renamed accordingly.
Note that in the documentation to checkNumberOfBondedInteractions(),
which was changed in a similar way, the argument `state` was
described as being the global state. Throughout the code, the local
state was used, however. This part of the documentation is now
changed, but if that was intended, then a follow-up change fixing
this might be necessary.
Change-Id: If913656fe9bf07567a29f68592402bb5ff17fc52
Pascal Merz [Tue, 9 Jul 2019 13:14:27 +0000 (07:14 -0600)]
Make relax_shell_flexcon independent of t_state
This change writes out the fields of t_state needed as explicit
arguments of relax_shell_flexcon instead of taking a pointer to
the full t_state object. This simplifies refactoring the t_state
object in future commits.
Change-Id: Iec3d0aea7462325963341b1d192cff697b764a7c
Kevin Boyd [Thu, 27 Jun 2019 02:38:22 +0000 (22:38 -0400)]
Clean up gmx solvate
Changed c-style struct to C++ with defaults, and changed a char*
field to a string.
Used vector instead of pointer array, and associated changes
to appending.
Moved a number of incrementing variables to local scope
Removed some unused variables and renamed confusing increment variables
Change-Id: I4a8def5621e133d48dc3dfb82efa916cbffabce6
Kevin Boyd [Fri, 5 Jul 2019 00:30:34 +0000 (20:30 -0400)]
Add implicit conversion of mdspan-containing types to mdspan.
Analagous to arrayRef function args taking vectors, arrays as input
Change-Id: I701f6373592cdf4d30649041ffe8dee76423433b
M. Eric Irrgang [Wed, 26 Jun 2019 18:25:20 +0000 (21:25 +0300)]
gmxapi Python package documentation layout.
Refs #2698
Change-Id: Id7fb8c6c91640660566394c37e63b5acb8bd9e3d
Pascal Merz [Tue, 2 Jul 2019 03:06:48 +0000 (21:06 -0600)]
Added trivial const qualifier to relax_shell_flexcon()
Change-Id: Id9e93c1254634e45fd0ef4f951a74670deae8607
Pascal Merz [Tue, 25 Jun 2019 19:34:44 +0000 (13:34 -0600)]
Const qualifier for forces in update_coords
Reflect the fact that forces are not changed in the coordinate
update routines.
Change-Id: I3c168bf185c6bb746c07d2bb10ff0bf6782c05cf
M. Eric Irrgang [Tue, 4 Jun 2019 21:12:58 +0000 (16:12 -0500)]
Incorporate sample_restraint repository.
This commit transfers development of the sample MD plug-in
code from an external repository to the GROMACS repository.
It should not be considered to add or change functionality,
but serves as a base for future integrated development.
The sample_restraint subtree serves as the root of a
template repository that a researcher could start from
to develop gmxapi MD extension code. It is meant to be
self-contained and built against an existing GROMACS
installation. Moving it to a GROMACS repo subdirectory
makes it less accessible to extension developers, but
easier to test and maintain for GROMACS developers. Near
term follow-up should include the creation of an external
skeleton repository, downstream subtree repository, or
"cookie cutter" repository drawing core content from the
GROMACS subtree.
* Import project from GitHub
- https://github.com/kassonlab/sample_restraint/releases/tag/v0.0.7.3
- Bundled sources for external projects are omitted.
- Bundled test data files are omitted.
* Updates for compatibility with the gmxapi Python package are deferred
to future changes.
Refs #3027
Change-Id: I1033a4a74dbce1bf1f9220a31cc9c15055c5dbaf
M. Eric Irrgang [Thu, 23 May 2019 21:44:00 +0000 (16:44 -0500)]
Begin CMake front-end for Python packaging machinery.
Allow gmxapi Python package sources to be discovered in the main GROMACS
CMake configuration step. During build, a functioning package will
appear in the build tree in
${CMAKE_BINARY_DIR}/python_packaging/src/gmxapi_staging that can be used
for tests and auto-generated documentation builds.
A Python 'sdist' can be built from this location in the build tree, or
the package can be built and installed directly from the source tree
with setup.py in python_packaging/src/
Refs: #2896
Change-Id: I225dd890c93405880ef057465eb7c159cd097665
Paul Bauer [Mon, 8 Jul 2019 09:20:13 +0000 (11:20 +0200)]
Fix postsubmit
Fix unstable post submit matrix by removing unused member functions in
constraint tests.
Change-Id: I156e8b1e6b1689afaa4269e2e4bb505768eea5c9
Szilárd Páll [Fri, 5 Jul 2019 21:09:36 +0000 (23:09 +0200)]
Eliminate a use of double precision in the GPU bonded kernels
The usual degree-to-radian conversion macro pulled in a double
literal.
Change-Id: I0d401427a2649e02dd81f603bf12979d7b15c583
Berk Hess [Thu, 14 Mar 2019 15:20:22 +0000 (16:20 +0100)]
Store nbnxm gridSet data in GridSetData
Change-Id: Ib018c59de965195c3093a0092eb711ff3d1ef41b
Mark Abraham [Sat, 6 Jul 2019 19:06:33 +0000 (21:06 +0200)]
Fix hwloc and sphinx detection
These were too noisy and were not implemented efficiently (e.g. not
caching the result of execute process). If an environment change is
needed to detect something (like loading a module) then we expect
developers to be able to unset a cache variable (or just regenerate a
new cache).
Refs #3024, #3011, #2998
Change-Id: I84dfb03856b9f900fb21004bc676e4ff2647a4b4
Mark Abraham [Tue, 2 Jul 2019 22:12:28 +0000 (00:12 +0200)]
Test specific hwloc versions
Now that releng understands the installed hwloc versions, we can
specify those in the matrix configurations. We can also specify that a
test configuration requires hwloc, but don't have to specify the
version. And we can leave it auto and provide a specific version of
the library to get detected automatically.
Refs #2899
Change-Id: I7f72b79746e77d04d9e2657b75821a8e3d62d83b
Kevin Boyd [Thu, 4 Jul 2019 22:17:27 +0000 (18:17 -0400)]
Change error message when attempting to offload to nonexistent GPUs
A user could interpret the error as no interactions detected rather
than no GPUs
Change-Id: Iedcc4555313edbb84a504bf01afc99c15eb4fdcb
Berk Hess [Mon, 24 Jun 2019 20:03:55 +0000 (22:03 +0200)]
Change affinity enum to enum class
Change-Id: I6accea15d82d2e2e911ce1e6f0735b7859b25021
Artem Zhmurov [Fri, 17 May 2019 14:08:24 +0000 (16:08 +0200)]
Remove PImpl scaffolding from CUDA version of LINCS
The CUDA implementation of LINCS was initially introduced as a
stand-alone feature. This required hiding CUDA-specific variables
and subroutines into the private implementation subclass. Since the
LINCS is not a part of Update and Constraints module, this is no
longer required and can be removed.
Refs #2816, #2888
Change-Id: I9698224d4702dfb8d99106999335c62e83a511df
Mark Abraham [Wed, 29 May 2019 13:37:21 +0000 (15:37 +0200)]
Improve tests of index block construction
Only trivial topologies were tested, and bugs have been
identified. Some test code is commented out until the fix lands.
Added missing Doxygen
Improved some assertions hit while investigating the issue.
Refs #2951
Change-Id: Ia03e38aca42be43efa1fd548c32e522d97e13e94
Mark Abraham [Wed, 3 Jul 2019 10:53:16 +0000 (12:53 +0200)]
Modernize FindHwloc.cmake
Users (and Jenkins) need the ability to direct finding a particular
hwloc version, e.g. from a module. This is particularly important now
that two API versions exist and are supported in GROMACS. Imported the
find module from Inria, which adds several useful capababilities,
including the ability to specify where to look for a specific version
of this library.
Removed the pkgconfig support from the imported FindHwloc.cmake
because it seems the hwloc.pc file from libhwloc does not lead to the
CMake idiomatic behaviour of HWLOC_LIBRARIES being the full path to
the library.
Refs #2539, #3011
Change-Id: I6837144a556f1c61a0f13406fb31fe1d527b65f4
M. Eric Irrgang [Thu, 23 May 2019 21:44:00 +0000 (16:44 -0500)]
Python bindings for mdrun.
This change makes a near-exact copy of the mdrun Python
bindings from gmxapi 0.0.7.
Reference https://github.com/kassonlab/gmxapi/releases/tag/v0.0.7.3
This establishes a base from which to transition Python bindings
development from GitHub to the GROMACS repository, so Python wrappers,
interface updates, and other changes are deferred to separate commits.
The provided interface largely opaque except for arguments passed the
same way as mdrun command line flags.
GROMACS must be built with GMXAPI=ON.
Fixes #2912
Change-Id: I2f51ae8f2369334d06a7ceaa8a4abf3ed9fc444a
Mark Abraham [Thu, 4 Jul 2019 08:58:38 +0000 (10:58 +0200)]
Further clang-tidy-7 fix
Change-Id: Ic2915148e6822f10fec6288f29aee55cddacf0f7
Szilárd Páll [Wed, 3 Jul 2019 12:33:24 +0000 (14:33 +0200)]
Bump latest CUDA + gcc in the pre-submit matrix
CUDA 10.1 and gcc 8 is the latest combination supported.
Change-Id: I4c7a5cc7817de1120e1d4d60783f00db70fb6e70
Szilárd Páll [Thu, 4 Jul 2019 10:01:19 +0000 (12:01 +0200)]
Fix a -Wformat-truncation warning in tng
Change-Id: Ib0cc008ad50957dededc857adf6bf921f98118b5
Artem Zhmurov [Thu, 27 Jun 2019 11:35:07 +0000 (13:35 +0200)]
Use reallocateDeviceBuffer(...) in CUDA version of SETTLE
Refs #2886 and #2888
Change-Id: Ia45254a24eda8e6ad151b1f4c6583b1a2c926004
Mark Abraham [Tue, 2 Jul 2019 21:24:18 +0000 (23:24 +0200)]
Fix clang-tidy-7 issues
These are detected on the mac build agent, but weren't on a Linux one,
e.g. because there is a system xdr library found on Linux.
Refs #2899
Change-Id: I591fda21c718b3a76f1fe17962150421f85fce22
Magnus Lundborg [Wed, 19 Jun 2019 11:28:42 +0000 (13:28 +0200)]
Vectorize GPU bonded data.
Loads data for bonds, angles, UB angles and pairs in blocks
of 3, 4, 4 and 3 ints respectively.
Refs: #2983
Change-Id: I37605395bd0c375b2eb4b94e76f9b30f6cb15e8d
Artem Zhmurov [Fri, 28 Jun 2019 14:52:17 +0000 (16:52 +0200)]
Early exit for CUDA versions of SETTLE and LINCS
Change-Id: Ia3c56a1be6b8ceaa8e1b7be956c1f1f3628c9139
Artem Zhmurov [Thu, 27 Jun 2019 10:04:23 +0000 (12:04 +0200)]
Memory management fixes in CUDA version of LINCS
This fix is to prepare LINCS to run with DD.
1. The masses array size depends on the current number of atoms
rather than on the number of constraints.
2. The size of other arrays should be based on the number of
threads launched on the GPU, which include padding added to
align coupled constraints with the thread blocks. Also
renamed variable according to conventions.
Refs #2885 and #2888
Change-Id: I20cb53ebc6da6a1ff2ee1e385613b27c4a01d11f
Mark Abraham [Tue, 2 Jul 2019 21:39:11 +0000 (23:39 +0200)]
Reorder options
If these are in alphabetical order, inserting new ones is less likely
to clash than if we always insert at the end of the list.
Change-Id: I3262959dc02deeaa628874028b10b9948f177842
Artem Zhmurov [Fri, 22 Mar 2019 15:47:54 +0000 (16:47 +0100)]
Combine CUDA Leap-Frog, LINCS and SETTLE. II.
Stand-alone CUDA implementations of Leap-Frog, LINCS
and SETTLE required additional scaffolding for integration
and testing. The most prominent part of this is the
management of coordinates, velocities and forces, which
is removed in this commit. Management of periodic boundary
conditions and virial reduction will be removed in
following commits.
Refs #2816, #2888
Change-Id: I4c65a6c7088fd8059f4e7fa3cb4637cb2af79ebc
Kevin Boyd [Tue, 2 Jul 2019 02:45:11 +0000 (22:45 -0400)]
Updates to math/vec.h
Removed unused functions
Added some file-level documentation indicating that the file
is deprecated and will be replaced in future
Change-Id: I2d6045884a4e14fa4278ef7a0606ccfbe2db7b63
Kevin Boyd [Fri, 18 Jan 2019 14:11:47 +0000 (09:11 -0500)]
Move more tools out of gmxana
Moved some files to simulationdatabase and updated references to them
refs #2727
Change-Id: Idf8375b672fbb6e4676fa22b9eb9c33d1b347166
M. Eric Irrgang [Thu, 16 May 2019 11:51:24 +0000 (14:51 +0300)]
Python detection consolidation.
* Move Python discovery up one level to the project root.
* Remove a TODO and clean up comments for validation test.
* Allow Sphinx discovery to appear in CMake output on first config.
Refs #2998
Change-Id: I16a406b2f21c517042729f3d970e0b5a79d40f33
M. Eric Irrgang [Wed, 19 Jun 2019 17:43:37 +0000 (20:43 +0300)]
Rearrange some definitions.
Reduce diff noise in related changes.
Change-Id: Ia1e30507918a3d2ecea7cc6d63b60312576077e9
Berk Hess [Mon, 24 Jun 2019 14:19:12 +0000 (16:19 +0200)]
Refactor thread affinity checking
Extracted the function detectDefaultAffinityMask() from
gmx_check_thread_affinity_set(). Now the affinity query check and
CPU count results are also reduced over MPI, so we avoid potential
deadlocks when results differ between ranks.
Change-Id: Ib382f47d63695357f8517b2657c3eb414e5afeef
Berk Hess [Mon, 24 Jun 2019 13:57:50 +0000 (15:57 +0200)]
Re-enable affinity diagonostics with MPI
Without thread-MPI, printing diagnostics of detected external
affinity settings when requesting affinities through GROMACS
got lost during refactoring. Now this is enabled again, which
also unifies code paths with and without thread-MPI.
Change-Id: I7a7e43ac4dc45713bda72d62b877d0c7108f1df7
Kevin Boyd [Tue, 18 Jun 2019 01:27:30 +0000 (21:27 -0400)]
Add mdspan basic elementwise math
BasicMatrix3x3 provides a replacement for the c-style tensors, but
has yet to replicate the tensor operations in math/vec.h. This provides
a subset of that functionality for msdpan in general, which can be used
for MultiDimArray trivially
Refs #2976
Change-Id: I17b77df032dbbfde0ff87108215edcec07fef6c4
M. Eric Irrgang [Wed, 26 Jun 2019 13:55:16 +0000 (16:55 +0300)]
Update gmxapi test fixtures.
* Rename spc216 to spc_water_box to be more clear and correct. Actual
solvent number depends on the results of 'solvate'.
* Improve error messaging (fix typo).
* Fix error in mdp parameter writing (surprised this didn't cause an
error previously...)
* Make spc_water_box fixture and other facilities from unit tests
available to integration tests.
* Use 2 instead of 1000 nsteps in test fixture TPR
Change-Id: I62eadeb5133d88ee11082755b1c33ae047d8f4dd
Kevin Boyd [Fri, 21 Jun 2019 01:25:42 +0000 (21:25 -0400)]
Add function for calculating 3x3 matrix determinant
C++ matrix compatible version of the det() function in math/vec.h
Change-Id: Iaa2749dc8059fb6c3c112c4798f8e8cc8eae993c
Kevin Boyd [Thu, 17 Jan 2019 03:54:11 +0000 (22:54 -0500)]
Remove references to ecutsGROUP
Changed warning that group scheme was deprecated to fatal error that
group scheme no longer exists
Modified warning when no cutoff scheme is specified
Change-Id: If971f88003af46c9ccedb98bb2d33987848509ed
Pascal Merz [Mon, 24 Jun 2019 22:37:35 +0000 (16:37 -0600)]
Fix module detection for physical validation
In I2edbe213, an implicit include() command was deleted, leading to
the module detection for physical validation not working anymore.
This change fixes the problem.
Refs #2615
Change-Id: If865227c2050f5739cd855261f6229f902fa0346
Szilárd Páll [Thu, 27 Jun 2019 17:53:08 +0000 (19:53 +0200)]
Refactor tracking of GPU short-range work/skipping
This change introduces a set of flags that, for each interaction
locality, whether there are short-range interactions computed and
exposes a query in the nonbonded module's API.
This allows consistent checks for both when work has been done
and whether results need to be reduced.
Refs #2986
Change-Id: I15020d83f73a132d9b8e93d7339529176396089a
Alan Gray [Wed, 12 Jun 2019 15:03:27 +0000 (08:03 -0700)]
Bug fix and simplification for CUDA X Buffer Ops
Fixes bug when there are more than two grids (ie they don't naively
map to local/nonlocal), by using a separate GPU memory space for
cxy_na and cxy_ind data for each grid (it previously only had one
local and one nonlocal space with overwriting). Also simpifies by now
only calling init fn once per NS step to set up data for all grids
(previously called for both local and nonlocal).
Change-Id: Ia2b97d22324aa97dca34b05da2eca2e2090372af
Berk Hess [Mon, 24 Jun 2019 09:50:35 +0000 (11:50 +0200)]
Fix MPI deadlock in affinity setting
Commit
96d28d6b introduced the possiblity for deadlocks when
the affinity masks changed over time.
Fixes #2989
Change-Id: I454b38deb9ff11c90cdf4a19aaa80e79e8898df5
Berk Hess [Wed, 19 Jun 2019 10:44:51 +0000 (12:44 +0200)]
Template free-energy kernel on differing coul/vdw soft-core
The power function used for the soft-core potential is expensive.
When using the same lambda and alpha parameters for Coulomb and VdW,
we can skip one power core, giving a 10% performance improvement.
Change-Id: I8733838c6c32ef2b6fee5a6fb97657679f9bd3b3
M. Eric Irrgang [Sat, 29 Jun 2019 06:08:43 +0000 (09:08 +0300)]
gmxapi Python package documentation infrastructure.
* Begin a documentation tree in python_packaging/documentation.
* Create a docker image for quick rebuild and review of Sphinx docs.
Refs #2698
Change-Id: Ib05e0309e60bfdadf52dadc7af0942c25c89f1e4
Szilárd Páll [Thu, 27 Jun 2019 15:26:50 +0000 (17:26 +0200)]
Create dedicated subcounter for nonbonded FEP
Now all nonbonded work has their own separate subcoutners which allows
measuring the performance of each task separately.
Refs #2997
Change-Id: I601445364592923d08087a858da4629b0b58ae76
Kevin Boyd [Sat, 12 Jan 2019 14:47:54 +0000 (09:47 -0500)]
Remove group scheme checks from runner
refs #1852
Change-Id: I906ffc7c063694fbbd17128ee0d1e62259040306
Magnus Lundborg [Tue, 18 Jun 2019 12:22:10 +0000 (14:22 +0200)]
Cleaning up the GPU bonded kernel
Improved variable names
Removed unnecessary includes
Refs #2988
Change-Id: Ife5f50e68f27994203305e376f2781a076052b64
Szilárd Páll [Mon, 13 Nov 2017 22:54:21 +0000 (23:54 +0100)]
Separate CPU NB kernel and buffer clearing subcounters
This is aimed to allow comparing the performance of the pair-interaction
kernels separately from the force buffer clearing.
Change-Id: Ifb2b4b3e5a43ac2ee547da651f9432a22fe58421
Christian Blau [Wed, 19 Jun 2019 11:35:18 +0000 (13:35 +0200)]
Add copy and move operations to density similarity
Enable copying and moving for DensitySimilarityMeasure.
Change-Id: Ie4250f67ae42b5cea227b7c17b0a932b04a42a63
Magnus Lundborg [Thu, 23 May 2019 14:39:50 +0000 (16:39 +0200)]
Fused GPU bonded kernels
To reduce the GPU kernel launch times the
GPU bonded kernels have been fused to a single kernel.
Refs #2818
Change-Id: I2299e1857df5db469739e3aefbc0f771968a6bd5
Christian Blau [Fri, 31 May 2019 06:30:39 +0000 (08:30 +0200)]
Add class to look up density fitting amplitudes
Adds a look-up for the weight used for spreading atoms on a grid and
calculate forces on them.
refs #2282
Change-Id: Iadd26f04265aa85ad30a827c92f2c5d5fc48ef85
Szilárd Páll [Thu, 27 Jun 2019 12:35:53 +0000 (14:35 +0200)]
Remove already resolved post-submit matrix TODO
Change-Id: I4b4f8c81376ac36c141e27fd6a3301f550968d1a
Kevin Boyd [Thu, 27 Jun 2019 01:57:48 +0000 (21:57 -0400)]
Use random water removal in gmx solvate
Given a number of solvent molecules to remove, gmx solvate
previously used a striding selection. This patch addresses a
TODO to use random selection
Change-Id: Icb065d61349a55b816337fa49470050eac194aa9