alexxy/gromacs.git
2 years agoSYCL PME Spread kernel
Andrey Alekseenko [Tue, 21 Sep 2021 18:54:59 +0000 (18:54 +0000)]
SYCL PME Spread kernel

2 years agoFix compilation with oneAPI 2021.2 and earlier
Andrey Alekseenko [Sat, 18 Sep 2021 21:30:45 +0000 (00:30 +0300)]
Fix compilation with oneAPI 2021.2 and earlier

We were using some PP defines introduced only in 2021.3. And, since they
are not standard, now we play it safe and don't use them if they are not
defined.

Bug introduced in MR !1888 (2861057f).

2 years agoFix rare UB / assertion failure in MdlibUnitTest
Andrey Alekseenko [Tue, 21 Sep 2021 12:21:15 +0000 (15:21 +0300)]
Fix rare UB / assertion failure in MdlibUnitTest

In LeapFrogHostTestRunner::integrate, we were checking whether
testData->mdAtoms_.ptype is null, but we were not initializing this
field.

This was leading to occasional assertion failure in
ArrayRef constructor from this pointer when begin+size was overflowing,
which caused end < begin.

Only unit tests were affected.

2 years agoFind MPI quietly on second pass
Mark Abraham [Tue, 21 Sep 2021 10:50:29 +0000 (12:50 +0200)]
Find MPI quietly on second pass

GROMACS expects package detection to be quiet upon repeat calls
to cmake, unless something has actually changed. It's fine to report
details thoroughly the first time through, however.

The previous code seemed to conflate verbosity with enforcing
requirements, which is now clearer.

2 years agoExpanded the GetIrTest behavior to enable efficient testing
Berk Hess [Tue, 21 Sep 2021 10:39:28 +0000 (10:39 +0000)]
Expanded the GetIrTest behavior to enable efficient testing

2 years agoPrepare ThreadedForceBuffer for FE kernel use
Berk Hess [Tue, 21 Sep 2021 07:52:15 +0000 (09:52 +0200)]
Prepare ThreadedForceBuffer for FE kernel use

Made the energy terms optional.
Skip reductions and allow nullptr/empty input for buffers that are
not used according to the stepwork flags.
Instantiate RVec versions of the templated types.

2 years agoAdd Ewald coefficients to nonbonded fep test
Berk Hess [Thu, 16 Sep 2021 20:37:54 +0000 (22:37 +0200)]
Add Ewald coefficients to nonbonded fep test

Refs #4151

2 years agoChange hpccm scripts to add heffte package
Gaurav Garg [Mon, 20 Sep 2021 09:12:26 +0000 (14:42 +0530)]
Change hpccm scripts to add heffte package

2 years agoSYCL: prepareGpuKernelArgument/launchGpuKernel
Andrey Alekseenko [Thu, 16 Sep 2021 16:36:00 +0000 (19:36 +0300)]
SYCL: prepareGpuKernelArgument/launchGpuKernel

Implement the functions for abstracting the kernel launch.

The abstraction is not idiomatic in SYCL, but it is conformant with what
is used in CUDA and OpenCL for PME kernel launches.

Refs #3927.

2 years agoFix Clang warnings for built-in XDR libraries
Andrey Alekseenko [Fri, 17 Sep 2021 07:14:19 +0000 (07:14 +0000)]
Fix Clang warnings for built-in XDR libraries

2 years agoMinor fixes to comments and Doxygen
Andrey Alekseenko [Tue, 14 Sep 2021 17:50:47 +0000 (20:50 +0300)]
Minor fixes to comments and Doxygen

- Fix issue number in cmake/gmxManageNvccConfig.cmake.
- Fix use of \tparam in CUDA code.

2 years agoMiscellaneous doxygen fixes
Artem Zhmurov [Thu, 16 Sep 2021 09:43:12 +0000 (12:43 +0300)]
Miscellaneous doxygen fixes

2 years agoUse a pointer to kernel parameters instead of a reference
Artem Zhmurov [Thu, 16 Sep 2021 09:18:34 +0000 (12:18 +0300)]
Use a pointer to kernel parameters instead of a reference

clang-tidy complains that the reference is not const.

Introduces in 98b7831420261dba9746da962551eb4376c77e3a

2 years agoEliminate GMX_MIMIC symbol from source code
Mark Abraham [Thu, 16 Sep 2021 10:45:17 +0000 (10:45 +0000)]
Eliminate GMX_MIMIC symbol from source code

2 years agoAdd SYCL compiler and MKL versions
Andrey Alekseenko [Thu, 16 Sep 2021 10:08:04 +0000 (10:08 +0000)]
Add SYCL compiler and MKL versions

2 years agoAdd SYCL implementation of LINCS
Artem Zhmurov [Wed, 15 Sep 2021 15:20:44 +0000 (15:20 +0000)]
Add SYCL implementation of LINCS

2 years agoMake static selections of large index groups work.
Mark Abraham [Wed, 15 Sep 2021 09:29:55 +0000 (11:29 +0200)]
Make static selections of large index groups work.

The default allocation size is the number of atoms, but this does not
work for large static selections. Such selections might well be used
with gmx distance for listing more pairs of atoms than there are
atoms. Fixed by allocating the needed size.

Fixes #4148

2 years agoFix clang warnings
Andrey Alekseenko [Wed, 15 Sep 2021 09:19:56 +0000 (09:19 +0000)]
Fix clang warnings

2 years agoMerge remote-tracking branch 'origin/release-2021' into master
Andrey Alekseenko [Tue, 14 Sep 2021 17:16:06 +0000 (20:16 +0300)]
Merge remote-tracking branch 'origin/release-2021' into master

Conflicts:
cmake/gmxVersionInfo.cmake
src/gromacs/ewald/pme_gather.cu
src/gromacs/ewald/pme_gpu_calculate_splines.cuh
src/gromacs/ewald/pme_gpu_internal.cpp
src/gromacs/ewald/tests/pmetestcommon.cpp
src/gromacs/gmxana/gmx_cluster.cpp
src/gromacs/gmxpreprocess/gen_vsite.cpp
src/gromacs/simd/support.cpp
src/gromacs/tools/trjconv.cpp

2 years agoAdd advanced event consumption accounting to GpuEventSynchronizer
Andrey Alekseenko [Tue, 14 Sep 2021 13:40:18 +0000 (13:40 +0000)]
Add advanced event consumption accounting to GpuEventSynchronizer

2 years agoReenable GPUs in CI
Paul Bauer [Tue, 14 Sep 2021 08:27:29 +0000 (08:27 +0000)]
Reenable GPUs in CI

2 years agoRevert "Temporarily lift requirement to detect GPUs in CI"
Paul Bauer [Mon, 7 Jun 2021 13:20:54 +0000 (15:20 +0200)]
Revert "Temporarily lift requirement to detect GPUs in CI"

This reverts commit b5f89c1771eda9912d94672bae31e611da9b4107.

Not needed after CI fix.

2 years agoAdded class for making QMMM-related topology modifications
Dmitry Morozov [Tue, 14 Sep 2021 07:57:27 +0000 (07:57 +0000)]
Added class for making QMMM-related topology modifications

2 years agoClang-format
Mark Abraham [Mon, 13 Sep 2021 16:19:37 +0000 (18:19 +0200)]
Clang-format

2 years agoSimplify MPI implementation checks.
M. Eric Irrgang [Mon, 13 Sep 2021 19:54:08 +0000 (19:54 +0000)]
Simplify MPI implementation checks.

2 years agoCalculate FEP (GPU) splines for all atoms.
Magnus Lundborg [Mon, 13 Sep 2021 18:08:15 +0000 (18:08 +0000)]
Calculate FEP (GPU) splines for all atoms.

Since charges can be 0 in one state, but not in the other, the splines
still need to be calculated. This was already ensured by c_skipNeutralAtoms
being false, but this commit makes that more clear.

Refs: #4139

2 years agoAdd release notes for transformation pull coordinate
Berk Hess [Tue, 7 Sep 2021 07:15:34 +0000 (09:15 +0200)]
Add release notes for transformation pull coordinate

2 years agoFix issues from downstream patches and suppressions
Mark Abraham [Fri, 10 Sep 2021 13:02:38 +0000 (15:02 +0200)]
Fix issues from downstream patches and suppressions

2 years agoAdded CP2K related options to CMake files
Dmitry Morozov [Mon, 13 Sep 2021 12:03:27 +0000 (12:03 +0000)]
Added CP2K related options to CMake files

2 years agoAllow flexible requirements for gmxManageMPI.cmake
M. Eric Irrgang [Mon, 13 Sep 2021 09:30:52 +0000 (09:30 +0000)]
Allow flexible requirements for gmxManageMPI.cmake

2 years agoRemove MPI_LINKER_FLAGS.
M. Eric Irrgang [Sun, 12 Sep 2021 12:34:18 +0000 (15:34 +0300)]
Remove MPI_LINKER_FLAGS.

GROMACS used MPI_LINKER_FLAGS before MPI_LINK_FLAGS
was standard, presumably. MPI_LINK_FLAGS has been superseded
by MPI_CXX_LINK_FLAGS.

Ref #3672

2 years agoAlways call `find_package(MPI)`.
M. Eric Irrgang [Sun, 12 Sep 2021 12:20:31 +0000 (15:20 +0300)]
Always call `find_package(MPI)`.

It is getting increasingly hard to coordinate MPI requirements across
the toolkit that a GROMACS or its CMake infrastructure may be involved
in.

This change allows is to call `find_package(MPI)` exactly once in the
GROMACS CMake configuration and begins to normalize infrastructure on
the modern CMake FindMPI.cmake module, which has been substantially
updated since the last major revisions to gmxManageMPI.cmake.

Also, try to decouple and compartmentalize GMX_MPI, GMX_LIB_MPI, and
the presence of MPI in the build system.

Follow-up changes can continue to refine the dependent use cases and
allow more coordination of the different MPI requirements in different
build configurations.

Refs #3672
Also refs #3776

2 years agoNormalize indentation.
M. Eric Irrgang [Sun, 12 Sep 2021 12:11:08 +0000 (15:11 +0300)]
Normalize indentation.

Make followup changes more readable and focused.

2 years agoFix GMX_VERSION_STRING_OF_FORK with REGRESSIONTEST_DOWNLOAD
Mark Abraham [Fri, 10 Sep 2021 13:26:25 +0000 (15:26 +0200)]
Fix GMX_VERSION_STRING_OF_FORK with REGRESSIONTEST_DOWNLOAD

Previously, the name of the fork was used to try to download
the normal regressiontests, which can't be found.

2 years agoCompute correct dV/dl when nstcalcenergy % nstdhdl != 0
Berk Hess [Fri, 10 Sep 2021 07:52:41 +0000 (07:52 +0000)]
Compute correct dV/dl when nstcalcenergy % nstdhdl != 0

2 years agoWord SIMD report better
Mark Abraham [Fri, 10 Sep 2021 07:30:46 +0000 (07:30 +0000)]
Word SIMD report better

Fixes #4135

2 years agoFix misc-misplaced-const clang-tidy warning
Andrey Alekseenko [Thu, 9 Sep 2021 14:46:43 +0000 (17:46 +0300)]
Fix misc-misplaced-const clang-tidy warning

The problem was introduced in 4a4ade89 (MR !1903).

As MPI_Comm is typedef'd as a pointer to a structure, declaring it const
would be of little use: the pointer itself would be const, not the data
it is pointing to.

2 years agoMinor improvements to SYCL utils and related clean-up
Andrey Alekseenko [Thu, 9 Sep 2021 20:43:06 +0000 (23:43 +0300)]
Minor improvements to SYCL utils and related clean-up

- ewald/pme_gpu_types_host.h: Remove unused forward declaration.
- gpu_utils/devicebuffer_sycl.h: Allow casting our "blackhole" class to
  nullptr.
- gpu_utils/sycl_kernel_utils.h: Make subGroupBarrier usable with any
  group dimension.

In preparation for #3927.

2 years agoFix typo in selectSpreadKernelPtr causing wrong kernel to be selected
Andrey Alekseenko [Thu, 9 Sep 2021 16:07:20 +0000 (16:07 +0000)]
Fix typo in selectSpreadKernelPtr causing wrong kernel to be selected

Fixes #4138

2 years agoExtract ThreadForceBuffer from listed forces
Berk Hess [Thu, 9 Sep 2021 15:08:44 +0000 (15:08 +0000)]
Extract ThreadForceBuffer from listed forces

2 years agoEnable GPU PME unit tests
Andrey Alekseenko [Thu, 9 Sep 2021 13:06:39 +0000 (13:06 +0000)]
Enable GPU PME unit tests

2 years agoRedesign GPU FFT abstraction
Gaurav Garg [Wed, 8 Sep 2021 18:40:06 +0000 (18:40 +0000)]
Redesign GPU FFT abstraction
- Modify interface to allow distributed FFT implementation in future
- Provide support for choosing FFT backend at runtime. E.g. CUFFT backend can be instantiated for single-GPU FFT but HeFFTe can be instantiated in case PME decomposition is used.

This is a pre-requisite for GPU PME-decomposition implementation.

Refs #3884

2 years agoIntroduce plumbing for ObservablesReducer
Mark Abraham [Tue, 7 Sep 2021 17:38:29 +0000 (17:38 +0000)]
Introduce plumbing for ObservablesReducer

2 years agoReplace ResidueTypeMap with std::unordered_map
Mark Abraham [Tue, 7 Sep 2021 05:22:49 +0000 (05:22 +0000)]
Replace ResidueTypeMap with std::unordered_map

2 years agoDescribe procedure and labeling for MR closure.
M. Eric Irrgang [Mon, 6 Sep 2021 11:26:15 +0000 (11:26 +0000)]
Describe procedure and labeling for MR closure.

2 years agoUse latest cmake for ASAN configuration
Mark Abraham [Mon, 6 Sep 2021 09:24:00 +0000 (09:24 +0000)]
Use latest cmake for ASAN configuration

2 years agoAdd support for transformation pull coordinates
Paul Bauer [Fri, 3 Sep 2021 11:16:31 +0000 (11:16 +0000)]
Add support for transformation pull coordinates

2 years agoMinor clean-up and optimization in DD topology
Berk Hess [Thu, 2 Sep 2021 16:09:11 +0000 (16:09 +0000)]
Minor clean-up and optimization in DD topology

2 years agoUse better data structure to implement ResidueType
Mark Abraham [Wed, 1 Sep 2021 07:39:10 +0000 (07:39 +0000)]
Use better data structure to implement ResidueType

Now insertions and lookups occur in constant time, rather than linear
time. This greatly improves the many lookups needed in grompp to
build the ndx file.

Renamed entry field as entries, which works better for naming a
container.

2 years agoRevert "Revert "Add muparser and CMake handling""
Paul Bauer [Tue, 31 Aug 2021 07:34:47 +0000 (07:34 +0000)]
Revert "Revert "Add muparser and CMake handling""

This reverts commit 4bfdcf2ccb21b5d0702da65083e52944ca112588.

We want to add this functionality again.

2 years agoFix grompp note on comm removal with posres
Berk Hess [Mon, 30 Aug 2021 08:19:53 +0000 (10:19 +0200)]
Fix grompp note on comm removal with posres

The grompp note with position restraints and center of mass motion
removal checked for absolute reference instead of COMM removal.
This bug was introduced in commit 0c7124d7 when fixing #3996.

Fixes #4128

2 years agoMake orires work with DD particle reordering
Berk Hess [Mon, 30 Aug 2021 07:53:39 +0000 (07:53 +0000)]
Make orires work with DD particle reordering

2 years agoGet rid of sycl::buffer::reinterpret
Andrey Alekseenko [Fri, 27 Aug 2021 12:12:57 +0000 (14:12 +0200)]
Get rid of sycl::buffer::reinterpret

This functionality is not properly supported in hipSYCL yet, and was
only needed in order to use atomic accessors. After fully switching to
atomic_ref, we can directly use buffers of Float3.

Closes #4063.

2 years agoSYCL: Fully switch to atomic_ref
Andrey Alekseenko [Sat, 28 Aug 2021 06:26:10 +0000 (06:26 +0000)]
SYCL: Fully switch to atomic_ref

2 years agoRemoves PME rank dependency on Non-local X H2D copy when buffer ops are on GPU but...
Gaurav Garg [Wed, 25 Aug 2021 08:56:18 +0000 (14:26 +0530)]
Removes PME rank dependency on Non-local X H2D copy when  buffer ops are on GPU but, Halo exchange happens over host-MPI.

This was not a problem until now because because !stepWork.haveGpuPmeOnThisRank will always evaluate to true in case of domain decomposition as PME will always be on a separate rank.
But now with PME decomposition this bug gets exposed.

Refs #4125

2 years agoDisable DLB if GPU direct communication Halo exchange is enabled.
Gaurav Garg [Fri, 27 Aug 2021 07:00:46 +0000 (07:00 +0000)]
Disable DLB if GPU direct communication Halo exchange is enabled.

2 years agoDocument Status labels for issues.
M. Eric Irrgang [Thu, 26 Aug 2021 22:13:51 +0000 (22:13 +0000)]
Document Status labels for issues.

2 years agoAdd missing Doxygen for SYCL functions and some others
Andrey Alekseenko [Tue, 24 Aug 2021 15:13:46 +0000 (18:13 +0300)]
Add missing Doxygen for SYCL functions and some others

Sometimes it was enough to convert existing comments to Doxygen.

Now there are no SYCL-related warnings when building webpage.

Also fixed a bit of trivial clang-tidy complaints.

2 years agoAdded toUpperCase and toLowerCase transformations of string
Dmitry Morozov [Thu, 26 Aug 2021 09:31:41 +0000 (12:31 +0300)]
Added toUpperCase and toLowerCase transformations of string

2 years agoMake the output of the constraints tests readable
Artem Zhmurov [Thu, 26 Aug 2021 15:42:19 +0000 (15:42 +0000)]
Make the output of the constraints tests readable

In parametrized tests, the failed runs output parameters. In order
for them to be meaningfull for human, the parameter types should be
provided with function that converts them to strings. This adds
these functions for t_pbc and ConstraintsTestSystem types in
constraints tests.

2 years agoAdd several more listed forces interaction types to nblib
Joe Jordan [Thu, 26 Aug 2021 07:14:30 +0000 (07:14 +0000)]
Add several more listed forces interaction types to nblib

2 years agoFind rocFFT for hipSYCL build when targeting AMD devices
Mark Abraham [Thu, 24 Jun 2021 06:52:54 +0000 (08:52 +0200)]
Find rocFFT for hipSYCL build when targeting AMD devices

We need this so that we can call rocFFT for PME GPU 3D FFT

Refs #3965

2 years agoFixed typos in replica exchange formulas
Carsten Kutzner [Wed, 25 Aug 2021 14:14:06 +0000 (16:14 +0200)]
Fixed typos in replica exchange formulas

2 years agoEnable atom reordering in WholeMoleculeTransform
Berk Hess [Wed, 25 Aug 2021 15:04:48 +0000 (15:04 +0000)]
Enable atom reordering in WholeMoleculeTransform

2 years agoAdd back original index file content that was accidentally overwritten by MR 1761
Carsten Kutzner [Wed, 25 Aug 2021 12:39:18 +0000 (14:39 +0200)]
Add back original index file content that was accidentally overwritten by MR 1761

Makes spc216.ndx functional again.

2 years agoUse better data structures in gmx chi
Mark Abraham [Thu, 29 Jul 2021 05:57:30 +0000 (07:57 +0200)]
Use better data structures in gmx chi

This removes part of the dependency of gmx chi histogramming from
ResidueType handling, which will permit changes to the latter that
will benefit both grompp and pdb2gmx

Replaced vectors that needed a stable numerical index to look up a
residue name with a map that uses the residue name to do the look up.

Removed capabilities of ResidueType that are no longer used.

2 years agoSimplify gmx chi
Mark Abraham [Fri, 20 Aug 2021 17:42:24 +0000 (17:42 +0000)]
Simplify gmx chi

This change is pure refactoring that prepares for performance
improvements to ResidueType handling that will benefit both grompp and
pdb2gmx.

Use vector and ArrayRef to replace C-style memory handling. Some
histogram vectors were being over-allocated by 1, which is no longer
safe to do now that the size of the vector is relevant when looping,
so those are reduced.

Eliminated and reduce scope of iteration variables. Removed an unused
function and some debug code in comments. Used const references rather
than pointers where possible. Used range-based for and algorithms in
some places that are now possible to do so.

2 years agoAdd tests for gmx chi
Mark Abraham [Fri, 20 Aug 2021 09:57:13 +0000 (09:57 +0000)]
Add tests for gmx chi

Fixed many memory leaks, unfortunately needing to add a new function
in order to work around the use of read_first_x().

Formally documented read_first_x() and read_next_x() as deprecated, as
the implementation of close_trx() in trxio.cpp has mentioned for a
long time.

2 years agoAvoid using cr->duty in setupStepWorkload
Szilárd Páll [Thu, 19 Aug 2021 20:16:20 +0000 (20:16 +0000)]
Avoid using cr->duty in setupStepWorkload

2 years agoRevert "Add workaround for CorrelationsTest with Intel ICPX"
Andrey Alekseenko [Thu, 19 Aug 2021 15:36:14 +0000 (18:36 +0300)]
Revert "Add workaround for CorrelationsTest with Intel ICPX"

This reverts commit 79fa11933f.

The proper fix was introduced in f3adc2fabe (MR !1794), and
then merged into master, making the workaround from 79fa11933f (MR
!1869) obsolete.

Refs #3955.

2 years agoUse workload flags more in do_md()
Szilárd Páll [Wed, 18 Aug 2021 19:09:05 +0000 (21:09 +0200)]
Use workload flags more in do_md()

2 years agoFix stale link in warning message.
Paul Bauer [Tue, 29 Jun 2021 15:03:40 +0000 (17:03 +0200)]
Fix stale link in warning message.

Pointed out by user, needs a matching fix in the regression tests.

2 years agoAdd separate PME rank SimluationWorkload flags
Szilárd Páll [Thu, 19 Aug 2021 13:16:27 +0000 (13:16 +0000)]
Add separate PME rank SimluationWorkload flags

Added the complete set of flags that indicate whether separate PME ranks
are used and what type of PP-PME communication is used.

These flags allow simplifications in the force schedule as well as reducing
the use of the confusingly defined cr->duty for now in do_force().

The additional benefit is a reduced reliance on passing around commrec
for checks related to parallelization.

Refs #3913

2 years agoUpdate Python package version to account for merge from gmxapi 0.2.2
M. Eric Irrgang [Wed, 11 Aug 2021 12:16:36 +0000 (15:16 +0300)]
Update Python package version to account for merge from gmxapi 0.2.2

2 years agoMerge remote-tracking branch 'origin/release-2021' into merge-2021-into-master
Andrey Alekseenko [Thu, 19 Aug 2021 09:51:13 +0000 (12:51 +0300)]
Merge remote-tracking branch 'origin/release-2021' into merge-2021-into-master

Resolved conflicts:
 - cmake/gmxVersionInfo.cmake
 - docs/CMakeLists.txt
 - python_packaging/src/gmxapi/export_system.cpp
 - python_packaging/src/setup.py
 - src/gromacs/domdec/domdec.cpp
 - src/gromacs/gmxana/gmx_chi.cpp

2 years agoRestrict scope of step and step_rel
Mark Abraham [Thu, 19 Aug 2021 05:45:25 +0000 (07:45 +0200)]
Restrict scope of step and step_rel

This prevents use-before-set issues

2 years agoClarify StepWorkload doxygen
Szilárd Páll [Wed, 18 Aug 2021 16:50:23 +0000 (18:50 +0200)]
Clarify StepWorkload doxygen

StepWorkload is rank-specific as it is also used on separate PME ranks.

Refs #3913

2 years agoAdd PP decomposition simluationWorkload flags
Szilárd Páll [Thu, 12 Aug 2021 16:57:42 +0000 (18:57 +0200)]
Add PP decomposition simluationWorkload flags

Add the complete set of flags that indicate whether there is PP
decomposition using CPU/GPU communication.

These flags allow simplifications in the force schedule.

The additional benefit is a reduced reliance on passing around commrec
for checks related to parallelization.

Refs #3913

2 years agoPrepares for the branch GROMACS 2021.4
Sebastian Wingbermuhle [Tue, 17 Aug 2021 13:47:33 +0000 (15:47 +0200)]
Prepares for the branch GROMACS 2021.4

2 years agoIntroduce ObservablesReducer
Mark Abraham [Wed, 18 Aug 2021 14:36:19 +0000 (14:36 +0000)]
Introduce ObservablesReducer

2 years agoFix compiler warnings introduced in 8f6ea9eb6c (!1865)
Andrey Alekseenko [Wed, 18 Aug 2021 09:42:11 +0000 (12:42 +0300)]
Fix compiler warnings introduced in 8f6ea9eb6c (!1865)

- Add ifdef around clang-specific pragmas.
- Add newline at the end of file.

2 years agoSimplify nblib listed forces type tests
Joe Jordan [Wed, 18 Aug 2021 09:31:18 +0000 (09:31 +0000)]
Simplify nblib listed forces type tests

2 years agoGpuEventSynchronizer: extract backend-specific functionality
Andrey Alekseenko [Wed, 18 Aug 2021 08:52:57 +0000 (08:52 +0000)]
GpuEventSynchronizer: extract backend-specific functionality

No new functionality or added checks.

Preparatory work for adding advanced event accounting logic to GpuEventSynchronizer.

DeviceEvent class contains only backend-specific functions and minimal sanity checks.
The more advanced logic (the one we're trying to fix in #3988) is kept in
GpuEventSynchronizer, and currently left unchanged (1:1 for OpenCL and SYCL,
very relaxed rules for CUDA).

Refs #2527, #3988.

2 years agoPrepares for the GROMACS 2021.3 release
Sebastian Wingbermuhle [Tue, 17 Aug 2021 13:26:33 +0000 (15:26 +0200)]
Prepares for the GROMACS 2021.3 release

2 years agoEnsure restart with update groups always works
Mark Abraham [Wed, 18 Aug 2021 08:04:59 +0000 (08:04 +0000)]
Ensure restart with update groups always works

2 years agoFix mdrun -ddorder pp_pme
Berk Hess [Wed, 18 Aug 2021 06:44:03 +0000 (06:44 +0000)]
Fix mdrun -ddorder pp_pme

2 years agoUpdate GPU CI configuration
Mark Abraham [Tue, 17 Aug 2021 17:01:37 +0000 (17:01 +0000)]
Update GPU CI configuration

CUDA 11.4.1 updates the most recent supported versions of gcc and
clang so we should update those also.

ROCM 4.3 is out, and is more likely than any older version to be the
one we will support in 2022, so we should update them.

Several recent bug fixes in hipSYCL are worth having to avoid needing
to work around them in other changes.

Previously we re-used the hipsycl container for clang-tidy with CUDA,
but as we wish that to remain at clang version 11, we need an alternative
solution. Simplest would be to add CUDA to the linting image, but that
also builds LLVM with TSAN. As that TSAN build didn't work out of the box
on a CUDA base image, I split the TSAN image off from the image that does
linting, and added CUDA to the latter.

2 years agoAdd workaround for CorrelationsTest with Intel ICPX
Andrey Alekseenko [Tue, 17 Aug 2021 16:39:44 +0000 (16:39 +0000)]
Add workaround for CorrelationsTest with Intel ICPX

2 years agoAdd basic tests for GpuEventSynchronizer
Andrey Alekseenko [Tue, 17 Aug 2021 15:29:14 +0000 (15:29 +0000)]
Add basic tests for GpuEventSynchronizer

2 years agoMove the ownership of the xUpdatedOnDevice event to update constraints
Artem Zhmurov [Tue, 17 Aug 2021 15:13:06 +0000 (15:13 +0000)]
Move the ownership of the xUpdatedOnDevice event to update constraints

2 years agoMake ExpfitTest.EffnERREST well defined
Andrey Alekseenko [Thu, 1 Jul 2021 10:39:50 +0000 (13:39 +0300)]
Make ExpfitTest.EffnERREST well defined

The original test was poorly constructed, and the τ₂ fitting parameter
was not defined for the test data: with α = 1, it could take any value
and the function would fit just the same.

Here, we replace the test data with the one where 0<α<1, and thus both
τ₁ and τ₂ are well-defined.

Also updated the documentation to reflect the actual function used in
the code: it was missing a prefactor of 2.

Closes #3955

2 years agoAdd Urey-Bradley listed interaction to nblib
Joe Jordan [Tue, 17 Aug 2021 09:03:09 +0000 (09:03 +0000)]
Add Urey-Bradley listed interaction to nblib

2 years agoRandom cosmetic fixes
Andrey Alekseenko [Tue, 17 Aug 2021 05:26:56 +0000 (05:26 +0000)]
Random cosmetic fixes

2 years agoAdd common header to GpuEventSynchronizer
Artem Zhmurov [Mon, 16 Aug 2021 16:04:30 +0000 (19:04 +0300)]
Add common header to GpuEventSynchronizer

Move the pre-processor conditional to a common header, that can be
included unconditionally. Will simplify code unification further on.

2 years agoUpdates for updated Python version requirement.
M. Eric Irrgang [Mon, 16 Aug 2021 16:01:38 +0000 (19:01 +0300)]
Updates for updated Python version requirement.

* Update documented required Python version.
* Remove some references to Python 3.6 and 3.7.
* Cite issues where appropriate.

Refs #3716

2 years agoAdd `breathe` package to CI Python venvs.
M. Eric Irrgang [Mon, 16 Aug 2021 10:08:01 +0000 (13:08 +0300)]
Add `breathe` package to CI Python venvs.

It seems likely that at least one of the projects documented in the
GROMACS manual will want to incorporate documentation from C++ headers
(Doxygen) into the Sphinx pages. We can update the container build
script in advance so that the package will be available for
experimentation and testing after the next container rebuild.

See also #2698

2 years agoRemove support for Python 3.6
Mark Abraham [Mon, 16 Aug 2021 12:05:49 +0000 (14:05 +0200)]
Remove support for Python 3.6

This version goes end-of-life before we will release GROAMCS 2022, so
we should stop testing it. We decided this a long time ago, but didn't
actually do it.

Refs #3708

2 years agogrompp no longer modifies nstcomm
Berk Hess [Tue, 29 Jun 2021 09:43:07 +0000 (11:43 +0200)]
grompp no longer modifies nstcomm

2 years agoUpdates to nblib listed forces kernels
Joe Jordan [Mon, 16 Aug 2021 14:18:22 +0000 (14:18 +0000)]
Updates to nblib listed forces kernels

2 years agoRework nblib kernel checks and setup code
Joe Jordan [Mon, 16 Aug 2021 13:49:41 +0000 (13:49 +0000)]
Rework nblib kernel checks and setup code