loadbalancing merge requestshttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests2022-03-08T11:00:40+01:00https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/22added maximum based tensor method2022-03-08T11:00:40+01:00Rene Halveradded maximum based tensor methodThe addition of the experimental maximum based tensor based load-balancing method.The addition of the experimental maximum based tensor based load-balancing method.Rene HalverRene Halverhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/21Fix point distance calculation2021-09-10T13:54:15+02:00Stephan SchulzFix point distance calculationThis fixes the point distance calculation of `ALL::Point`s.
See #60 for more info.This fixes the point distance calculation of `ALL::Point`s.
See #60 for more info.Release v1.0Stephan SchulzStephan Schulzhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/20update the outline used for VTK output when balance is called2021-07-27T15:10:55+02:00Rene Halverupdate the outline used for VTK output when balance is calledSolution to issue #56 by updating the outlines after calling balance()Solution to issue #56 by updating the outlines after calling balance()https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/19Integrate vtk92021-07-27T11:04:49+02:00Stephan SchulzIntegrate vtk9Integrate the VTK 9 patches and generate a suitable docker container to test.
TODO:
- upload container to registry
- add gitlab-ci entry to run itIntegrate the VTK 9 patches and generate a suitable docker container to test.
TODO:
- upload container to registry
- add gitlab-ci entry to run ithttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/18Fortran error tests2021-02-10T13:43:34+01:00Stephan SchulzFortran error testsFortran error tests should be added, as suggested in #13.
For this some things have to be reworked, additional error checking enabled (such as correct permissions for already existing VTK output directories), that we can use to test aga...Fortran error tests should be added, as suggested in #13.
For this some things have to be reworked, additional error checking enabled (such as correct permissions for already existing VTK output directories), that we can use to test against.Release v1.0Stephan SchulzStephan Schulzhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/17Remove badges generation from CI2021-02-01T14:02:56+01:00Stephan SchulzRemove badges generation from CIRemoves generation of badges, which we are not using (or actually generating really).Removes generation of badges, which we are not using (or actually generating really).Release v1.0Stephan SchulzStephan Schulzhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/16Add example projects to CI2021-01-27T13:23:02+01:00Stephan SchulzAdd example projects to CIAlso run example projects during CI.
This way we can find problems with external usage.Also run example projects during CI.
This way we can find problems with external usage.Release v1.0Stephan SchulzStephan Schulzhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/15CMake projectconfig2021-01-25T17:49:55+01:00Stephan SchulzCMake projectconfigCreate a CMake package, so the library can be found via `find_package`.
Also create an example CMake project, that uses the library.Create a CMake package, so the library can be found via `find_package`.
Also create an example CMake project, that uses the library.Release v1.0https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/14Release test/0.8.02020-12-04T11:45:19+01:00Stephan SchulzRelease test/0.8.0Testing the automated creation of generic packages to simplify releasing.Testing the automated creation of generic packages to simplify releasing.Release v0.9https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/8Refactor2020-12-03T08:49:22+01:00Rene HalverRefactorUpdate the master branch to provide a working CI again and small changes since last merge from branch 'refactor'Update the master branch to provide a working CI again and small changes since last merge from branch 'refactor'Release v0.9Rene HalverRene Halverhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/11Feature/simple test cases2020-12-01T11:43:42+01:00Stephan SchulzFeature/simple test casesAdd a few simple examples, that can also be used as test cases of "full" programs.
This should test quite a few cases for all methods.
Current important cases to consider:
* [x] 1) Staggered: 1d static system must obey minimum domain...Add a few simple examples, that can also be used as test cases of "full" programs.
This should test quite a few cases for all methods.
Current important cases to consider:
* [x] 1) Staggered: 1d static system must obey minimum domain size constraint
* [ ] 2) Staggered: 1d system, which shrinks, but must obey minimum domain size constraint (will fail at the moment)
* [x] 3) Staggered: same as 1), but from Fortran to show that interface as well.
1) is given in `ALL_Staggered.cpp` already.
Also todo is to create a way to check the tests for validity. The proposal here is to generate a known good output and commit that. Then run the example and compare its output with the known good solution. For this we need a method to compare (with a certain degree of leeway to account for floating point rounding) the two sets of data. This shold not be difficult to code, once the output format is decided upon.https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/10Feature/fix warnings2020-10-28T11:05:53+01:00Stephan SchulzFeature/fix warningsWe should be able to compile using `-Wall -Wextra -Werror`.
Currently the easy fixes are done, but in the Unstructured load balancer many variables are declared (and set), but not used. I assume, this is the case due to the in progress ...We should be able to compile using `-Wall -Wextra -Werror`.
Currently the easy fixes are done, but in the Unstructured load balancer many variables are declared (and set), but not used. I assume, this is the case due to the in progress state of the method.
Other open problems are:
* unused parameters for some implementations of `setAdditionalData`. If these declaration have to be there we may silence them with proper annotations (for GCC: `__attribute__((unused))`[(ref)](https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#Common-Variable-Attributes), which should be put into a compiler dependant preprocessor switch to also set this for clang etc. as well as for readability. Maybe `known_unused` is a good keyword)
* ignored const on function return (`ALL_LB.hpp:163: const int getNVertices() { return vertices.size(); }`)
This branch should only be merged to `refactor` after !9 has been merged!.https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/9introduce namespace ALL, ALL_ -> ALL::2020-10-13T15:44:39+02:00Stephan Schulzintroduce namespace ALL, ALL_ -> ALL::This is API breaking (technically), but replacing `ALL_` with `ALL::` and `ALL<` with `ALL::ALL<` in the calling code should fix all problems (see the changes to the example programs).
This corresponds to #3.This is API breaking (technically), but replacing `ALL_` with `ALL::` and `ALL<` with `ALL::ALL<` in the calling code should fix all problems (see the changes to the example programs).
This corresponds to #3.https://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/7Allowing reuse of vertices between steps and bugfixes2019-11-12T18:04:30+01:00Rene HalverAllowing reuse of vertices between steps and bugfixesAllow the reuse of the vertices between load-balancing steps without the need to update by calling set_vertices() before balance().Allow the reuse of the vertices between load-balancing steps without the need to update by calling set_vertices() before balance().Rene HalverRene Halverhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/4staggered fixes2019-06-04T17:29:23+02:00Steffen Secklerstaggered fixesFixes staggered grid:
- set_min_domain_size
- fix memory leaks
Fixes destructor for ALL:
- delete the balancer in ALL using the load balancer type.
- fix memory leaksFixes staggered grid:
- set_min_domain_size
- fix memory leaks
Fixes destructor for ALL:
- delete the balancer in ALL using the load balancer type.
- fix memory leaksRene HalverRene Halverhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/3Voronoi2019-02-13T09:56:11+01:00Rene HalverVoronoiVoronoi method works, no need for separate branch anymore.Voronoi method works, no need for separate branch anymore.Rene HalverRene Halverhttps://gitlab.jsc.fz-juelich.de/SLMS/loadbalancing/-/merge_requests/2Unstructured2018-12-11T10:09:48+01:00Rene HalverUnstructuredUnstructured development in a state, that further development is bug fixing and performance improvement, which can be done in the master branch, as the feature itself is workingUnstructured development in a state, that further development is bug fixing and performance improvement, which can be done in the master branch, as the feature itself is workingRene HalverRene Halver