alexxy/gromacs.git
3 years agoRedevelopment of GPU Force Reduction/Buffer Ops
Alan Gray [Tue, 29 Sep 2020 11:27:18 +0000 (11:27 +0000)]
Redevelopment of GPU Force Reduction/Buffer Ops

Introduces a new purpose-build class for GPU force reduction, which
replaces the previous force buffer ops mechanism.

Refs #3370

3 years agoClarifications for ResourceAssignment interface.
M. Eric Irrgang [Mon, 28 Sep 2020 18:11:28 +0000 (21:11 +0300)]
Clarifications for ResourceAssignment interface.

Add and update some documentation. Improve error messages. Improve
file naming.

Refs #3644

3 years agoSupport ResourceAssignment interface for tMPI library builds.
M. Eric Irrgang [Sat, 26 Sep 2020 15:05:25 +0000 (18:05 +0300)]
Support ResourceAssignment interface for tMPI library builds.

Fixes #3644

3 years agoAllow MPI-enabled clients to provide the library communicator.
M. Eric Irrgang [Mon, 28 Sep 2020 12:55:50 +0000 (15:55 +0300)]
Allow MPI-enabled clients to provide the library communicator.

Provide library helpers and template headers to allow clients to
createContext() with explitly assigned resources.

Relates to #3644

3 years agoUse MpiContextManager to provide MPI communicator.
M. Eric Irrgang [Mon, 28 Sep 2020 12:35:43 +0000 (15:35 +0300)]
Use MpiContextManager to provide MPI communicator.

* Give MpiContextManager a MPI_Comm and make it movable so that it can
  be provided while creating the ContextImpl.
* In gmxapi::ContextImpl::launch(), use an MpiContextManager instance as
  the source for the SimulationContext communicator.

Refs #3644

3 years agoRelocate MpiContextManager.
M. Eric Irrgang [Mon, 28 Sep 2020 12:24:03 +0000 (15:24 +0300)]
Relocate MpiContextManager.

* Change the scope of the MpiContextManager instance from the Session
  to the ContextImpl.
* Tighten up some details to help maintain clarity over subsequent
  changes.

Refs #3644

3 years agoInterface updates allowing client to provide MPI communicator.
M. Eric Irrgang [Sat, 26 Sep 2020 09:54:32 +0000 (12:54 +0300)]
Interface updates allowing client to provide MPI communicator.

gmxapi::MpiContextManager is acquired earlier in API initialization
and has a larger role in setting up the SimulationContext as a member
of gmxapi::ContextImpl. The MpiContextManager can be acquired with
appropriate default values for the various MPI-related GROMACS build
variants, or it can be initialized with user-provided resources.

gmxapi::Context creation method has been updated to accept optional
non-default resources from the client.

A template header helps MPI-enabled client code to provide resources
to createContext(). The client requires minimal awareness of whether
GROMACS was built for thread-MPI or an installed MPI library.
Thread-MPI details are isolated from any translation unit that might
include the client MPI tool chain.

Refs #3644

3 years agoRemove single-dimension limitation from GPU halo exchange
Alan Gray [Mon, 28 Sep 2020 20:53:07 +0000 (20:53 +0000)]
Remove single-dimension limitation from GPU halo exchange

Allows GPU halo exchange to be active when the number of dimensions is
greater than 1, thus simplifying the codepath logic.

Partly addresses #3370

3 years agoAdd nblib backend: Part 1 of 2
Joe Jordan [Mon, 28 Sep 2020 17:21:12 +0000 (17:21 +0000)]
Add nblib backend: Part 1 of 2

3 years agoFix error in ensureReferenceCount
Andrey Alekseenko [Mon, 28 Sep 2020 15:35:39 +0000 (15:35 +0000)]
Fix error in ensureReferenceCount

We were overwriting input value and comparing it to itself.

3 years agoAdd initial nblib public API
Joe Jordan [Mon, 28 Sep 2020 10:35:59 +0000 (10:35 +0000)]
Add initial nblib public API

3 years agoFixed incorrect enum
Giovanni Bussi [Mon, 28 Sep 2020 07:46:02 +0000 (07:46 +0000)]
Fixed incorrect enum

Harmless, since both have value 1, but confusing

3 years agoadd detection of CPX 5300 w/ 2FMAs
Artem Zhmurov [Mon, 28 Sep 2020 06:00:40 +0000 (06:00 +0000)]
add detection of CPX 5300 w/ 2FMAs

Signed-off-by: Hammond, Jeff R <jeff.r.hammond@intel.com>
3 years agoCompare trajectories to reference data
Pascal Merz [Sun, 27 Sep 2020 11:23:20 +0000 (11:23 +0000)]
Compare trajectories to reference data

This introduces functionality to compare trajectories (x/v/f/box)
to reference data. It also allows to define the maximum number of
frames which should be checked. The latter allows to have a single
reference data, but decide based on the simulation setup (e.g.
parallelization scheme) how many frames are expected to match
before numerical divergence gets too large. To allow this, the
reference data functionality needs to give the option to ignore
unused reference frames.

3 years agoRoll back "Ensure the modular simulator runs only when it should"
Pascal Merz [Sun, 27 Sep 2020 06:55:11 +0000 (00:55 -0600)]
Roll back "Ensure the modular simulator runs only when it should"

The fact that GMX_USE_MODULAR_SIMULATOR does not lead to an error
when the run is not supported is a feature, not a bug. It allows
to run the entire test set using modular simulator wherever is is
implemented by using "export GMX_USE_MODULAR_SIMULATOR=ON; make check",
even if modular simulator is disabled by default. It is the only way
to make sure that modular simulator is a valid replacement for the
legacy code path.

Similarly, using "export GMX_DISABLE_MODULAR_SIMULATOR=OFF; make check"
allows to check the legacy code path, even if modular simulator is used
by default.

3 years agoFix #3668
Andrey Alekseenko [Fri, 25 Sep 2020 20:41:08 +0000 (20:41 +0000)]
Fix #3668

Following the implementation in Google's cpu_features tool made by an
Intel engineer used for original implementation, we only check for the
presence of the second AVX-512 FMA unit if the CPU supports AVX-512F
(foundation), which should be present for all AVX-512 implementations.

That's distinct from the way cpu_features tool does AVX detection, but
should be no less reliable.

3 years agoReplace FramesToCompare enum by MaxNumFrames
Pascal Merz [Fri, 25 Sep 2020 16:37:48 +0000 (16:37 +0000)]
Replace FramesToCompare enum by MaxNumFrames

This allows to explicitly chose a maximum number of frames to be
tested, rather than only chosing to compare the first frame or
all frames. It also implements to chose the number of frames for
tests comparing to reference data rather than only when comparing
simulation results.

3 years agoEnsure the modular simulator runs only when it should
Mark Abraham [Fri, 25 Sep 2020 15:21:43 +0000 (17:21 +0200)]
Ensure the modular simulator runs only when it should

Previous checks were ineffective if an unsupported inputrec
was accompanied by GMX_USE_MODULAR_SIMULATOR.

Change-Id: I74e4c177b6b4d7dc25bb3ca32a7dff5a6027276e

3 years agoFix bad logic in regressiontest branch selection
Paul Bauer [Fri, 25 Sep 2020 13:15:44 +0000 (15:15 +0200)]
Fix bad logic in regressiontest branch selection

Change-Id: I1eed8a5228362d34af9426a73c2a9d06eae7a714

3 years agoAllow deployment of nightly webpage to server
Paul Bauer [Thu, 24 Sep 2020 14:46:18 +0000 (14:46 +0000)]
Allow deployment of nightly webpage to server

Still needs to have variables defined on Gitlab.

Change-Id: Id7378942555b69df19cbe377d7c90851d6cf56ba

3 years agoImprove OpenCL log/error reporting
Szilárd Páll [Thu, 24 Sep 2020 12:44:09 +0000 (12:44 +0000)]
Improve OpenCL log/error reporting

Improve reporting by making messages more specific; also added logging of binary caching also for the case where a cache files is found in addition to the case where no cache file is present

3 years agoFix false-positives in checkDeviceBuffer for OpenCL
Andrey Alekseenko [Wed, 23 Sep 2020 16:34:29 +0000 (18:34 +0200)]
Fix false-positives in checkDeviceBuffer for OpenCL

The function takes required number of elements as its argument, but then
compares it with the buffer size measured in bytes.

3 years agoReport name of missing OpenCL kernels
Mark Abraham [Wed, 23 Sep 2020 15:26:35 +0000 (17:26 +0200)]
Report name of missing OpenCL kernels

Refs #3686

3 years agoAccess write checkpoint functionality directly
Pascal Merz [Wed, 23 Sep 2020 16:59:06 +0000 (16:59 +0000)]
Access write checkpoint functionality directly

3 years agoImplement Berendsen thermostat for modular simulator
Pascal Merz [Tue, 8 Sep 2020 03:03:50 +0000 (21:03 -0600)]
Implement Berendsen thermostat for modular simulator

Refs #3423

3 years agoPermit comparison of only first frames
Mark Abraham [Wed, 23 Sep 2020 14:56:57 +0000 (14:56 +0000)]
Permit comparison of only first frames

Integration tests of mdrun sometimes want to compare only the
quantities from the first frame of a trajectory.  This is now
permitted.

3 years agoAdd regression tests for gpucomm paths
Alan Gray [Wed, 23 Sep 2020 07:02:47 +0000 (07:02 +0000)]
Add regression tests for gpucomm paths

Adds regression test runs with GPU communication codepaths enabled for
both thread-MPI and MPI builds.

3 years agoSplit mdrun-test
Mark Abraham [Tue, 22 Sep 2020 13:18:17 +0000 (13:18 +0000)]
Split mdrun-test

The PME tests are the most time-consuming part, so have been split off.

3 years agoDraft: Fix compilation
Andrey Alekseenko [Tue, 22 Sep 2020 12:50:41 +0000 (12:50 +0000)]
Draft: Fix compilation

Also, make ctor etc always defined for DeviceInformation. Otherwise it's
just ugly.

3 years agoMake Oneapi use CCache again
Paul Bauer [Tue, 22 Sep 2020 10:04:35 +0000 (10:04 +0000)]
Make Oneapi use CCache again

The custom before_script meant that builds are not using the cache.

Change-Id: Iaf0546e2dd48cf2a6438c2aa0d6461977ee6b2dd

3 years agoNelder-Mead downhill simplex optimisation
Christian Blau [Tue, 22 Sep 2020 07:30:09 +0000 (07:30 +0000)]
Nelder-Mead downhill simplex optimisation

Nelder-Mead simplex optimiser, according to Saša Singer and
John Nelder (2009), Scholarpedia, 4(7):2928. doi:10.4249/scholarpedia.2928

Implementation for functions returning real values from
vectors of reals.

3 years agoCheckpointdata DD followup
Pascal Merz [Mon, 21 Sep 2020 16:34:37 +0000 (16:34 +0000)]
Checkpointdata DD followup

3 years agoWrong logic was used
Mark Abraham [Mon, 21 Sep 2020 14:21:14 +0000 (16:21 +0200)]
Wrong logic was used

Log files in gitlab gave the warning message for twin-cutoff, but we
are not setting these env variables.

3 years agosimd/impl_none: add missing macros
Gilles Gouaillardet [Mon, 21 Sep 2020 15:01:19 +0000 (15:01 +0000)]
simd/impl_none: add missing macros

add macros for GMX_SIMD_HAVE_HSIMD_UTIL_{FLOAT,DOUBLE}

3 years agoFix script building all CI containers and add gcovr
Paul Bauer [Mon, 21 Sep 2020 13:33:09 +0000 (13:33 +0000)]
Fix script building all CI containers and add gcovr

Additions in 48f9056c196fda338b83e74212d4f9bc5ab3474e broke the script
by changing the default values, while
e9cb08d17b06bc5649d316ee06159b633db9b3d8 inadvertently made all images
use the oneapi images as base.

Also add gcovr tool to documentation images, preparing for coverage
builds.

Change-Id: I0eb9a3bf1e8b45dd4b4a3a41fd214678c4c1c4c2

3 years agoAdd OpenSSH to docs build images
Paul Bauer [Mon, 21 Sep 2020 08:44:58 +0000 (10:44 +0200)]
Add OpenSSH to docs build images

Change-Id: I9b1994a78ef869f3147aa45b5893645ce74f5e04

3 years agoAWH avoid constructors only differing in parameter types
Magnus Lundborg [Mon, 21 Sep 2020 09:07:13 +0000 (09:07 +0000)]
AWH avoid constructors only differing in parameter types

Also clarified a function name in the tests.
Tests had to be updated since the force constant (now removed)
should not have been used before at all.

3 years agoDo not run PinnedMemoryCheckerTest when there are no CUDA-capable devices
Artem Zhmurov [Mon, 21 Sep 2020 08:26:52 +0000 (08:26 +0000)]
Do not run PinnedMemoryCheckerTest when there are no CUDA-capable devices

If there are no CUDA devices in the system, cudaPointerGetAttributes(...)
CUDA API call returns an error. This causes the GPU Utils tests to fail
with fatal error.

Fixes #3679

3 years agoMove the message to the deprEcation-policy file
Artem Zhmurov [Mon, 21 Sep 2020 07:59:07 +0000 (07:59 +0000)]
Move the message to the deprEcation-policy file

Also update the message according to the move

3 years agoAdd NVIDIA GPU compatibility check, following issue #3125
Andrey Alekseenko [Sun, 20 Sep 2020 09:39:41 +0000 (09:39 +0000)]
Add NVIDIA GPU compatibility check, following issue #3125

The incompatibility is documented, but better to verify.

3 years agoSet active device when the device stream manager is initialized
Artem Zhmurov [Sun, 20 Sep 2020 09:18:16 +0000 (09:18 +0000)]
Set active device when the device stream manager is initialized

Replace the direct CUDA API call in PME by the wrapper called
when the device context and streams are created.

3 years agoAllow using old environment variables, but print deprecation message
Artem Zhmurov [Sun, 20 Sep 2020 08:56:29 +0000 (08:56 +0000)]
Allow using old environment variables, but print deprecation message

The following environment variables were unified in
2c644d3b5ebb64f6c89a5db36160c9dc30329888:

GMX_CUDA_NB_ANA_EWALD and GMX_OCL_NB_ANA_EWALD into GMX_GPU_NB_ANA_EWALD
GMX_CUDA_NB_TAB_EWALD and GMX_OCL_NB_TAB_EWALD into GMX_GPU_NB_TAB_EWALD
GMX_CUDA_NB_EWALD_TWINCUT and GMX_OCL_NB_EWALD_TWINCUT into GMX_GPU_NB_EWALD_TWINCUT

This allows user to use old environment variables, but prints deprecation message.

3 years agoRefactor v-rescale thermostat to allow for other algorithms
Pascal Merz [Tue, 8 Sep 2020 02:40:22 +0000 (20:40 -0600)]
Refactor v-rescale thermostat to allow for other algorithms

This is pure refactoring, splitting the element class (interfaces with
the simulator and other objects) from the actual implementation of the
thermostat. This will allow to implement other thermostat algorithms
using the same infrastructure.

Refs #3423

3 years agoImprove naming and docs.
M. Eric Irrgang [Tue, 15 Sep 2020 19:18:22 +0000 (22:18 +0300)]
Improve naming and docs.

Follow up from https://gitlab.com/gromacs/gromacs/-/merge_requests/541

Fixes #3673

3 years agoRemove cutoff_scheme from t_forcerec and interaction_const_t
Berk Hess [Fri, 18 Sep 2020 09:01:21 +0000 (09:01 +0000)]
Remove cutoff_scheme from t_forcerec and interaction_const_t

3 years agoDocument a Mdrunner::BuilderImplementation member.
M. Eric Irrgang [Mon, 14 Sep 2020 16:15:04 +0000 (19:15 +0300)]
Document a Mdrunner::BuilderImplementation member.

Fixes #3665.

3 years agoMove bNonbonded flag from t_forcerec to SimulationWorkload
Berk Hess [Thu, 17 Sep 2020 12:13:57 +0000 (14:13 +0200)]
Move bNonbonded flag from t_forcerec to SimulationWorkload

The flag to store whether nonbonded calculations are requested to be
skipped (set by env.var.) is move from t_forcerec to
SimulationWorkload. A minor side-effect of this change is that the
nbnxn module logs kernels choices when this option is set.

3 years agoRename VRescaleThermostat to VelocityScalingTemperatureCoupling
Pascal Merz [Thu, 3 Sep 2020 03:56:11 +0000 (21:56 -0600)]
Rename VRescaleThermostat to VelocityScalingTemperatureCoupling

Also moves vrescalethermostat.h/cpp to
velocityscalingtemperaturecoupling.h/cpp.

This reflects upcoming changes making the thermostat more general.
As these changes touch a lot of lines in the implementation file,
renaming the file at the same time throws git diff off. Doing the
renaming in a separate commit should therefore make for easier reviewing.

Refs #3423

3 years agoRemove the CPU hardware context from the general list of test hardware contexts
Artem Zhmurov [Thu, 17 Sep 2020 12:49:10 +0000 (12:49 +0000)]
Remove the CPU hardware context from the general list of test hardware contexts

Having CPU code path in the general hardware context list leads to having
many if (CPU) clauses. This complicates things in places, where there are
several CPU runners. Since in most cases (i.e. everywhere apart from Ewald
tests) the runners know the hardware they should use, the selection of CPU
or GPU is now done when the runners are created. TestHardwareContext is
to become TestDeviceContext and will only hold GPU devices.

3 years agoRemove unnecessary condition in DD force buffer resizing
Berk Hess [Wed, 16 Sep 2020 13:33:12 +0000 (15:33 +0200)]
Remove unnecessary condition in DD force buffer resizing

Removed a condition that was no longer necessary, as exactly
the same condition is checked inside the resize() method of
ForceHelperBuffers.

3 years agoActivate AMD OpenCL build in release matrix
Mark Abraham [Wed, 16 Sep 2020 15:41:05 +0000 (15:41 +0000)]
Activate AMD OpenCL build in release matrix

This had never been turned on.

Fixed missing file for OpenCL installation.

Also reordered the install list in alphabetical order, per style.

Refs #3272

3 years agoReduce transitive dependencies in arrayref.h
Sebastian Keller [Tue, 15 Sep 2020 09:09:45 +0000 (11:09 +0200)]
Reduce transitive dependencies in arrayref.h

Reduce transitive dependencies on other GROMACS headers.
* Replace `GMX_ASSERT` with `assert`
* Replace `index` with `difference_type`

Refs #3674

3 years agoExtend user docs on affine transformations for densities
Christian Blau [Tue, 15 Sep 2020 14:55:31 +0000 (16:55 +0200)]
Extend user docs on affine transformations for densities

Add more extensive information to the reference manual and .mdp option description
in the user-guide.

Fix one documentation typo.

3 years agoImplement modular checkpointing
Pascal Merz [Wed, 16 Sep 2020 07:34:44 +0000 (07:34 +0000)]
Implement modular checkpointing

Using the CheckpointData format introduced in a parent commit, this
rewrites checkpointing for the modular simulator to completely use
the new format.

The CheckpointHelper is now passing a CheckpointData object to its
clients (instead of a legacy t_state object). Clients are now stored
in a map, as they are identified by their unique key to be able to
assign the correct CheckpointData sub-objects at reading and writing.

If checkpoint reading occured, the newly introduced
CheckpointHelperBuilder receives the CheckpointData object read at the
runner level from the ModularSimulator. It then initializes its clients
with their respective, read-only CheckpointData subobjects.

The ICheckpointHelperClient interface is adapted to reflect above
changes.

The ModularSimulatorAlgorithmBuilder is slightly simplified thanks to
to the introduction of a proper builder for the CheckpointHelper.

The ComputeGlobalsElement is simplified, as it is not required to know
about the needs of communication of the EnergyData object which
depends on checkpoint reading.

Finally, all elements which are checkpoint clients are updated to
implement the new design. Note that they all introduce their own
checkpoint versioning, as the data being checkpointed is opaque to the
checkpointing infrastructure.

Closes #3517
Closes #3422
In partial fulfillment of #3419

3 years agoUse common device context header for OpenCL
Artem Zhmurov [Tue, 15 Sep 2020 16:43:43 +0000 (16:43 +0000)]
Use common device context header for OpenCL

This moves the OpenCL specific definition into under macro in the main header.
Makes device stream and context definitions more consistent.

3 years agoBreak off some tests from mdrun tests
Paul Bauer [Tue, 15 Sep 2020 14:17:42 +0000 (14:17 +0000)]
Break off some tests from mdrun tests

Done to not hit the time limit in CI with OpenCL.

Change-Id: Ic2e2686530aef51475ab7f49c1bfd29442c3fc0a

3 years agoRemove outdated workaround.
M. Eric Irrgang [Fri, 11 Sep 2020 13:52:31 +0000 (16:52 +0300)]
Remove outdated workaround.

Remove a workaround left over from before issue #2689 was closed.

Refs #2689

3 years agoOnly extend Ewald table when necessary
Berk Hess [Tue, 15 Sep 2020 09:40:23 +0000 (11:40 +0200)]
Only extend Ewald table when necessary

The Ewald correction table for non-bonded interactions was always
extended by change 24595dbb. No this is limited to free-energy
runs only.

3 years agoEnable GMX_GPU=SYCL compiles
Erik Lindahl [Tue, 15 Sep 2020 12:38:19 +0000 (12:38 +0000)]
Enable GMX_GPU=SYCL compiles

Update CMake config to properly apply -fsycl, but only to files
including actual SYCL code, and get linking working.  Tested
with both dpcpp and icpx, and added trivial SYCL code to a
dimmy device context, just to ensure compiles and linking works
both with and without SYCL support enabled.

Also fixes a number of new warnings about unused variables now
detected with dpcpp and icpx.

3 years agoFix missing installed files for OpenCL PME
Paul Bauer [Tue, 15 Sep 2020 11:31:21 +0000 (13:31 +0200)]
Fix missing installed files for OpenCL PME

Change fa6862b510acd7b0225580b661446c278dbb0039 forgot to modify
the CMake files to install the renamed OpenCL header files.

Refs #3664

3 years agoAll coordinates in the density-guided-simulation group undergo an affine transformati...
Christian Blau [Tue, 15 Sep 2020 11:21:14 +0000 (11:21 +0000)]
All coordinates in the density-guided-simulation group undergo an affine transformation before density-fitting forces on them are evaluated, which allows rotation, transformation and scaling of coordinates.

Resolves part of #3581.

3 years agoFix task assignment with multiruns
Berk Hess [Tue, 15 Sep 2020 10:17:46 +0000 (10:17 +0000)]
Fix task assignment with multiruns

Change e0f481ae broke the physical node hardware detection and task
assignment by passing the simulation instead of the world communicator
to the creation of the physical node communicator. Now the world
communicator is stored and passed.

Fixes #3666

3 years agoFix minor warnings noticed by Intel compilers
Erik Lindahl [Tue, 15 Sep 2020 07:59:04 +0000 (09:59 +0200)]
Fix minor warnings noticed by Intel compilers

New things found by dpcpp & ipcx compilers.

3 years agoPass devFlags into createSimulationWorkload replacing bools
Alan Gray [Fri, 11 Sep 2020 11:51:58 +0000 (04:51 -0700)]
Pass devFlags into createSimulationWorkload replacing bools

3 years agoFix VSite out of range error message
Artem Zhmurov [Tue, 15 Sep 2020 05:46:44 +0000 (08:46 +0300)]
Fix VSite out of range error message

3 years agoCreate and use SimulationInput module.
M. Eric Irrgang [Mon, 14 Sep 2020 15:37:40 +0000 (15:37 +0000)]
Create and use SimulationInput module.

Allow Mdrunner to accept a handle to SimulationInput from the client.

This change adds an input method to the MdRunner builder to provide
access to resources provided through the new SimulationInput module.

Refs #3374

3 years agoUse common device context header for OpenCL
Artem Zhmurov [Mon, 14 Sep 2020 09:46:19 +0000 (09:46 +0000)]
Use common device context header for OpenCL

This moves the OpenCL specific definition into under macro in the main header.
Makes device stream and context definitions more consistent.

3 years agoRefine the CMakeLists.txt for MPI gmxapi tests.
M. Eric Irrgang [Fri, 11 Sep 2020 13:47:56 +0000 (16:47 +0300)]
Refine the CMakeLists.txt for MPI gmxapi tests.

Exclude tests when GMX_MPI is not set. Normalize test working directory.

3 years agoDecouple nsttcouple and nstpcouple from nstlist
Berk Hess [Sun, 13 Sep 2020 09:30:20 +0000 (09:30 +0000)]
Decouple nsttcouple and nstpcouple from nstlist

The default values of nsttcouple and nstpcouple of -1 would cause
nstlist to be used as the coupling interval. But coupling intervals
should not be linked to nstlist. The default is now set to 10, which
will not change settings for mdp files where nstlist was not specified
as the default nstlist value is also 10.

3 years agoFix GPU device info broadcast with MPI
Mark Abraham [Sat, 12 Sep 2020 11:42:42 +0000 (13:42 +0200)]
Fix GPU device info broadcast with MPI

3 years agoMove src/api cmake logic into src/api cmake file
ejjordan [Fri, 11 Sep 2020 12:34:27 +0000 (14:34 +0200)]
Move src/api cmake logic into src/api cmake file

3 years agoRemove leftover data after refactoring
Paul Bauer [Fri, 11 Sep 2020 06:26:02 +0000 (08:26 +0200)]
Remove leftover data after refactoring

The modular simulator still hat the old datastructure for initial
lambdas, causing a complaint in the CI.

Change-Id: Ic97ee36ebea8cd40d7c513f44f767751f5c271cc

3 years agoFix doxygen for unused parameters.
Paul Bauer [Fri, 11 Sep 2020 10:11:17 +0000 (10:11 +0000)]
Fix doxygen for unused parameters.

Recent addition to the code didn't use the parameters for the SYCL
stubs, leading to errors from doxygen.

Change-Id: I61666e4e76cfdcfa5c618c7bc6f58cf24ea2b4f8

3 years agoDefine stubs for GPU data structures in SYCL to make SYCL compilation possible
Artem Zhmurov [Fri, 11 Sep 2020 09:06:16 +0000 (09:06 +0000)]
Define stubs for GPU data structures in SYCL to make SYCL compilation possible

3 years agoForce use of Python3 in Doxygen CMakeFiles
Paul Bauer [Fri, 11 Sep 2020 07:58:35 +0000 (07:58 +0000)]
Force use of Python3 in Doxygen CMakeFiles

Change-Id: Ia7d430fef243d6b3236de903907dbf8e9ba92222

3 years agoSimplify and fix how lambda values are set
Christian Blau [Thu, 10 Sep 2020 19:04:49 +0000 (19:04 +0000)]
Simplify and fix how lambda values are set

Simplified setCurrentLambdasLocal method to return an array of
current lambda points. Removed the faulty use of lam0 as initial
lambdas.

3 years agoFix Visual Studio build
Roland Schulz [Thu, 10 Sep 2020 04:51:17 +0000 (21:51 -0700)]
Fix Visual Studio build

- Add /permissive- flag to disable MSVC extension. Fixes
  ternary operator.
- Fix usage of %l for size_t
- Disable non-wroking FFTW SIMD check
- Missing imports
- Compiler warning fix

3 years agoSimpify CheckpointData
Roland Schulz [Wed, 9 Sep 2020 20:06:04 +0000 (13:06 -0700)]
Simpify CheckpointData

3 years agoMinor fix to the printout for regtest builds
Paul Bauer [Thu, 10 Sep 2020 12:37:44 +0000 (14:37 +0200)]
Minor fix to the printout for regtest builds

Was not printing the branch it is building the archive from.

Change-Id: I597e68a3780d3f300b5a4be447ecac7834126408

3 years agoUse common device context header for OpenCL
Artem Zhmurov [Thu, 10 Sep 2020 13:49:23 +0000 (13:49 +0000)]
Use common device context header for OpenCL

This moves the OpenCL specific definition into under macro in the main header.
Makes device stream and context definitions more consistent.

3 years agoAdd angles to ListedForces selections
Berk Hess [Thu, 10 Sep 2020 11:37:55 +0000 (13:37 +0200)]
Add angles to ListedForces selections

3 years agoUse device information object instead of id when performing device checks
Artem Zhmurov [Thu, 10 Sep 2020 10:27:03 +0000 (10:27 +0000)]
Use device information object instead of id when performing device checks

Initializing plain object is dangerous, since it then can be used to set
active device, with default id. This change ensures that proper device
will be used, even if the high level device management is used in the
checks.

3 years agoFix memory corruption with strings in grompp
Berk Hess [Thu, 10 Sep 2020 09:00:52 +0000 (11:00 +0200)]
Fix memory corruption with strings in grompp

Using a string of more than 31 characters as a default value for
a string entry in the mdp file would lead to memory corruption.

3 years agoReplace GMX_ASSERT with assert in vectypes.h
Joe Jordan [Thu, 10 Sep 2020 06:42:13 +0000 (06:42 +0000)]
Replace GMX_ASSERT with assert in vectypes.h

It is inappropriate to have a wrapper around assert in a public API.
This change removes dependence on GMX_ASSERT in one component of the
installed API. Transient dependencies related to inclusion of
gmxassert.h in vectypes.h are also removed.

3 years agoUnify more functions in CUDA and OpenCL implementations of NBNXM
Artem Zhmurov [Thu, 10 Sep 2020 05:39:44 +0000 (05:39 +0000)]
Unify more functions in CUDA and OpenCL implementations of NBNXM

3 years agoExplicitly include STL headers used in test.
M. Eric Irrgang [Wed, 9 Sep 2020 17:34:13 +0000 (17:34 +0000)]
Explicitly include STL headers used in test.

Include `array` and `optional` in strconvert.cpp test.

Fixes #3657

3 years agoRestore GMX_FORCE_UPDATE_DEFAULT_GPU environment varible check
Artem Zhmurov [Wed, 9 Sep 2020 06:29:11 +0000 (09:29 +0300)]
Restore GMX_FORCE_UPDATE_DEFAULT_GPU environment varible check

This line was accidently removed during unrelated refactoring
in 75c17a76b69c1ade1f29fdce9fe2f6dd78221c1d .

3 years agoBug fix in call to deviceStreamManager constructor
Alan Gray [Tue, 8 Sep 2020 16:34:25 +0000 (16:34 +0000)]
Bug fix in call to deviceStreamManager constructor

DeviceStreamManager construction expects
(deviceInfo,useGpuForPme,havePpDomainDecomposition,
doGpuPmePpTransfer,useGpuForUpdate,useTiming) but was being passed
(deviceInfo,useGpuForPme,useGpuForNonbonded,
havePPDomainDecomposition,useGpuForUpdate,useTiming) Now passing
correct arguments through simulationWork. Also ensures update stream
is correctly set up when gpu buffer ops are enabled.

Fixes #3639 and #3641

3 years agoFix clang-tidy warning
Andrey Alekseenko [Tue, 8 Sep 2020 12:51:19 +0000 (14:51 +0200)]
Fix clang-tidy warning

3 years agoFix "GPU detection deactivated" message when it in fact failed
Andrey Alekseenko [Tue, 8 Sep 2020 12:29:08 +0000 (14:29 +0200)]
Fix "GPU detection deactivated" message when it in fact failed

Also remove duplicate GPU detection check, and raise log level on failure from GPU detection API.

3 years agoMake legacy energy elements use CheckpointData
Pascal Merz [Tue, 8 Sep 2020 13:20:07 +0000 (13:20 +0000)]
Make legacy energy elements use CheckpointData

This enables some legacy energy elements to write to CheckpointData,
namely

* energyhistory_t
* delta_h_history_t
* ekinstate_t

Refs #3517
Refs #3422
Refs #3419

3 years agoAllow density-guided simulations to translate structure
Christian Blau [Tue, 8 Sep 2020 12:49:24 +0000 (12:49 +0000)]
Allow density-guided simulations to translate structure

Density-guided simulations acquire a new feature that allows a
user-defined shift on the structure before forces are calculated.

First step to address #3581

3 years agoRemove dependency of EnergyData on t_state
Pascal Merz [Tue, 8 Sep 2020 10:56:09 +0000 (10:56 +0000)]
Remove dependency of EnergyData on t_state

EnergyData used a t_state object to calculate the conserved energy.
This responsibility is now given to the respective temperature /
pressure control algorithms. This allows to remove the dependency
of EnergyData on t_state.

This also introduces the conserved energy as a test quantity in the
tests comparing legacy and modular simulator code paths.

Refs #3419

3 years agoMove densityfitting to its own directory
Paul Bauer [Tue, 8 Sep 2020 07:29:07 +0000 (07:29 +0000)]
Move densityfitting to its own directory

Preparing for adding more applied_forces modules

Change-Id: Ieec20d1f0752a369d26352ec6b0997812decd21e

3 years agoFix output string for GPU support
Paul Bauer [Mon, 7 Sep 2020 22:47:17 +0000 (22:47 +0000)]
Fix output string for GPU support

This used to print only disabled or CUDA because the variable was only 0
or 1.

Change-Id: Ie406a36e61617c29629f97d811eb93244b5a2e3a

3 years agoFix clang-format failure in coordinatetransformation.h
Christian Blau [Mon, 7 Sep 2020 16:12:52 +0000 (18:12 +0200)]
Fix clang-format failure in coordinatetransformation.h

Fixes formatting clang format error.

3 years agoMake EnergyOutput::addDataAtEnergyStep independent of t_state
Pascal Merz [Mon, 7 Sep 2020 20:23:36 +0000 (20:23 +0000)]
Make EnergyOutput::addDataAtEnergyStep independent of t_state

EnergyOutput::addDataAtEnergyStep took a pointer to a full t_state,
although it only uses a small part of the object. This makes the
dependencies explicit. This is a necessary step to make EnergyData
independent of t_state.

Refs #3419

3 years agoRemove in-body comment from init_forcerec(...) function
Artem Zhmurov [Mon, 7 Sep 2020 10:58:02 +0000 (13:58 +0300)]
Remove in-body comment from init_forcerec(...) function

Fixes doxygen warning

3 years agoFix clang-format
Artem Zhmurov [Mon, 7 Sep 2020 17:44:03 +0000 (20:44 +0300)]
Fix clang-format

3 years agoReplace filename-based initialization functions.
M. Eric Irrgang [Mon, 7 Sep 2020 16:13:31 +0000 (16:13 +0000)]
Replace filename-based initialization functions.

Decouple Mdrunner from the details of `-s` and `-cpi`
filename arguments when establishing initial simulation state.

Declare gmx::SimulationInput and the library utilities needed to
replace direct reference to TPR and CPT files in Mdrunner.
Use SimulationInput and accompanying utilities to remove read_tpx_state
and load_checkpoint from Mdrunner.

Allows TPR and checkpoint file handling to be considered separately
from Mdrunner initialization in future development.

Refs #3374