Project

General

Profile

2021-infrastructure-stable

Due in about 6 months (08/15/2020)

changes to versions of things supported and tested should stabilize at least several weeks before the beta

11%

91 issues   (10 closed — 81 open)

Issues by
Bug

1/12
Feature

7/22
Task

2/57
Related issues
Bug #3113: Use of read_tps_conf is depecrated
Bug #3136: gmxapi.operation data flow topology unclear or incomplete
Bug #3141: gmxapi File placeholders missing from beta release
Bug #3150: gmxapi data type annotations are confusing and inadequate
Bug #3197: Rework usage of gitattributes and filter in source code formatting scripts
Bug #3228: gmxapi.commandline_operation should gracefully handle input/output file arguments
Bug #3246: GPU code misses settle error check, simulation crashes with segfault without any further output
Bug #3273: gmxapi setup.py complains too much
Bug #3274: Remove ARM NEON SIMD
Bug #3282: Copyright message is not compatible with the clang-format
Bug #3284: gmx msd cannot output two xvg files as suggested by command line option
Bug #3333: Assertion failure when trying to get unit test help
Feature #2816: GPU offload / optimization for update&constraits, buffer ops and multi-gpu communication
Feature #2817: GPU X/F buffer ops
Feature #2885: CUDA version of LINCS
Feature #2886: CUDA version of SETTLE
Feature #2888: CUDA Update and Constraints module
Feature #2994: Data flow topology in gmxapi 2020
Feature #2996: gmxapi execution model
Feature #3029: GPU force buffer ops + reduction
Feature #3032: Clean up dev-manual structure
Feature #3038: Improvements to MD plugin development environment
Feature #3117: Coulomb FEP PME on GPU
Feature #3140: Allow explicit input definition for gmxapi.operation function wrapper
Feature #3142: centralize and clarify GPU force buffer clearing
Feature #3147: gmxapi workflow checkpointing
Feature #3148: Roadmap for gmxapi filesystem interactions.
Feature #3152: Infrastructure and patterns for expressing public interfaces
Feature #3160: implement direct comm for different src/target memory spaces
Feature #3363: Improve FEP testing
Feature #3379: C++ API for simulation input and output
Feature #3393: Allow gmxapi.commandline_operation to use STDIN
Feature #3394: improve gmxapi error handling
Task #2395: break up commrec
Task #2396: Refactor trajectoryanalysis module to allow handling of trajectory frame manipulation
Task #2481: Update gmx report-methods to write more method information
Task #2698: gmxapi documentation integration
Task #2699: Test POWER9
Task #2706: Rework classic QM/MM interface
Task #2795: Incorporate regressiontests into core gromacs
Task #2833: Update topology datastructures
Task #2875: SIMD version of the free-energy kernel
Task #2916: Decide future of symtab
Task #2925: BasicVector addition operator yields unexpected result when adding scalar
Task #2998: Update Python detection
Task #2999: Move all logging in core library and tools to use MDLogger
Task #3010: upgrade OpenCL stack on bs-gpu01
Task #3013: Extend logging levels in MDLogger
Task #3014: gmxapi example Python scripts
Task #3015: Create central log handle for calls to gmx_fatal
Task #3035: Provide central logger in wrapper binary
Task #3072: Use gmx::index for indexing instead of int
Task #3076: Selectively activate MdModules during simulation instead of instantiating all of them
Task #3093: rework GPU direct halo-exchange related force reduction complexities
Task #3096: Multiple time-stepping: correct coupling algorithms, virial calculation and energy output
Task #3130: Interim handling of gmxapi data references.
Task #3131: support ccache with clang-tidy
Task #3133: Cookiecutter for sample_restraint
Task #3153: Let CMake process module directories earlier to support more modern CMake idioms.
Task #3155: Unify DD and CPU-GPU gather & scatter
Task #3156: move ddUsesGpuDirectCommunication and related conditionals into the workload data structures
Task #3162: Add virtual site support to GPU version of update-constraints.
Task #3171: schedule CPU H2D force contribution in separate stream
Task #3180: Remove extra D2H and H2D copies of coordinates when the COM motion is compensated.
Task #3183: enable bonded interactions on GPU
Task #3189: implement heuristics for switching between different spread/gather kernel layouts
Task #3203: Tools should check that input is valid when first using it
Task #3220: change rolling pruning scheduling with GPU update
Task #3229: Move checkpointing to the beginning of the MD step
Task #3235: Run mdrun-test in parallel and make sure they tests test critical quantities like energy conservation
Task #3236: GPU update-constraints: Pulling support
Task #3258: GPU update-constraints: Enable free energy perturbation
Task #3260: keep up to with mdspan reference implementation
Task #3264: Break up MdrunMpiTests
Task #3271: Adopt PEP-518 for Python package build system configuration.
Task #3297: Require gcc > 5
Task #3308: GPU update-constraints: Cycle counting for GPU launches
Task #3337: GPU update-constraints: Merge Leap-Frog, LINCS and SETTLE kernels.
Task #3338: GPU update-constraints: Do not save intermediate coordinates if there is no constraints
Task #3339: GPU LINCS: Warp-level synchronization for coupled constraints
Task #3340: GPU LINCS: Move more data to local/shared memory
Task #3341: GPU LINCS: Use analytical solution for the matrix A inversion.
Task #3342: GPU SETTLE: Read only one index per water molecule.
Task #3343: GPU SETTLE: Use different ordering for matrices.
Task #3344: GPU update-constraints: Better virial reduction.
Task #3345: GPU update-constraints: Reconsider the naming of coordinates buffer in constraints.
Task #3346: GPU SETTLE: Use the same parameters and parameters initialization in GPU and CPU versions of SETTLE.
Task #3348: GPU LINCS and SETTLE: unify virial reduction
Task #3349: GPU update-constraints: Unify the PBC management.
Task #3392: Unify box correction for modular simulator