Release 2022.07.18
Upgrade instructions
From #4033 (Add a global enum class Phase):
If you have defined an enum class Phase
in a metavariables, instead #include "Parallel/Phase.hpp"
and replace the enum definition with using Phase = Parallel::Phase;
From #4046 (Use global phase to simplify some code):
- In a PhaseAction, the first template parameter should be removed.
- Also restrict phase names to those in Parallel::Phase
From #4044 (Interpolation without Interpolator ParallelComponent for time-dependent maps.):
No changes needed for the user, but anyone changing the interpolator code will see differences:
- intrp::Vars::PointInfoTag now has a different type.
- InterpolationTargetVarsFromElement and some of the functions in InterpolationTarget_detail now take different arguments.
From #4050 (Add ResourceInfo tags to DgElementArray and IntrpTarget):
If your executable has the DgElementArray parallel component, add in the following block to the input file
ResourceInfo:
AvoidGlobalProc0: false
If your executable has at least one InterpolationTarget parallel component, add the following under the ResourceInfo
block in the input file
Singletons:
TargetName:
Proc: Auto
Exclusive: false
From #4056 (Allow CCE singletons to be placed on specific cores):
If you have CCE in your executable, add the following block to your input file:
ResourceInfo:
Singletons:
CharacteristicEvolution:
Proc: Auto
Exclusive: false
BoundaryComponentName:
Proc: Auto
Exclusive: false
where BoundaryComponentName
is one of AnalyticWorldtubeBoundary
, H5WorldtubeBoundary
, or GhWorldtubeBoundary
depending on what your boundary component is.
From #4023 (Remove CERROR, specialize error macros for compile time and runtime):
This requires updaying computing resources to use >= gcc-9 or >= clang-9
From #4055 (Finish transition to using Parallel::Phase):
- Always use
Parallel::Phase
as the type of a phase - Do not specify a
determine_next_phase
member function for metavariables. Instead define thedefault_phase_order
member variable
From #4086 (Consolidate generic parallel actions):
If you used any actions in the eliminated directories, you will need to update the appropriate include statement.
If you added any actions to those directories, move them into ParallelAlgorithms/Actions
From #4099 (Combine registration phases):
If you are using Parallel::Phase::RegisterWithObserver, instead use Parallel::Phase::Register
From #4107 (Turn on LTS in BBH):
If using the BBH executable, update the time stepping options to match those of the test input file.
Merged pull-requests (65)
New features (3):
- Add a global enum class Phase (#4033)
- Add support for Charm++ projections (#4053)
- Add Translation control system (#3992)
General changes (54):
- Update ci dox (#4020)
- Use global phase to simplify some code (#4046)
- Relax more HwenoImpl test tolerances (#4039)
- Avoid dynamic allocations for single-point Variables data (#4035)
- Silence unused-value warning in clang 14 (#4048)
- Store element name in ElementVolumeData (#4042)
- Simplify Overloader and remove make_overloader (#4045)
- Change protocol helpers to normal metafunctions (#4049)
- Explicitly use interpolator source vars in intrp framework (#4014)
- Update Expanse build to use Charm 7 and Intel MPI (#3968)
- Add to_basis&to_quad, cleanup IO test helpers (#4043)
- Remove explicit exceptions from Python h5::File wrapper (#4041)
- Add ResourceInfo tags to DgElementArray and IntrpTarget (#4050)
- Increase timeout in RotScaleTrans test (#4065)
- Remove buffer from Ccz4::upper_spatial_z4_constraint (#4059)
- Fix Ccz4 temporary expression calculation (#4060)
- Forward declare Ccz4 temporary tags (#4061)
- Allow CCE singletons to be placed on specific cores (#4056)
- Add fallback reconstructor types to finite difference (#4057)
- Update domain in BBH ID input file (#3926)
- Add get_data_by_element function to VolumeData (#4054)
- Add to ParallelComponent section of Parallelization DevGuide (#4051)
- Use smaller random values for sin argument (#4067)
- Remove CERROR, specialize error macros for compile time and runtime (#4023)
- Enable TensorExpressions to internally split up equations (#3842)
- Finish transition to using Parallel::Phase (#4055)
- Split up and organize H5 tests (#4068)
- Enable Wcns5z reconstruction to use adaptive fallback (#4058)
- Rename algorithm impl to distributed object (#4069)
- Subcell TCI improvements, step 1 (#4012)
- Hybrid EOS (#3481)
- Shorten rotation control system test (#4074)
- Increase GH constraints timeout (#4076)
- Add Brill Lindquist analytic data (#4075)
- Remove specific enthalpy (#4080)
- Generalize control system test helper (#4071)
- Add Expansion and Rotation control systems to BBH exec (#4016)
- Add tags and initialization action for AMR flags (#3092)
- Consolidate generic parallel actions (#4086)
- Add TensorExpression support for assigning to doubles (#4079)
- Support for reading single and double precision HDF5 volume data (#4084)
- Add SphericalKerrSchild (#4073)
- Add xsimd support (#4081)
- Add temperature to equations of state (#4085)
- Disallow antisymmetries in TensorExpressions (#4090)
- Store number of indices in Tensor class (#4094)
- Add dependencies for SphericalKerrSchild executable (#4092)
- Add Noah Ring to metadata (#4101)
- Run a checkpoint test in its own directory (#4098)
- Add David Wu to Metadata (#4100)
- Increase tolerance for Eos test helper (#4104)
- Combine registration phases (#4099)
- Turn on LTS in BBH (#4107)
- Add build type to info at compile time (#4116)
Bugfixes (8):
- Remove load balancing test (#4052)
- Cleanup execs and libs after running tests (#4040)
- Interpolation without Interpolator ParallelComponent for time-dependent maps. (#4044)
- Fix Charm v7.0.0 AlgorithmArray.def.h patch (#4063)
- Clean up register phase in BBH exec (#4083)
- Fix non-positive times in get_data_by_element (#4106)
- Pin Spack version on CI (#4110)
- Minor fixes in quickstart guide, fix rendering of $ in docs (#4095)
Contributors (10): @nilsdeppe, @kidder, @wthrowe, @knelli2, @markscheel, @macedo22, @yoonso0-0, @nilsvu, @DavidWu421, @nring21