waLBerla merge requestshttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests2023-07-28T11:10:18+02:00https://i10git.cs.fau.de/walberla/walberla/-/merge_requests/623[FIX] AMD Usage2023-07-28T11:10:18+02:00Markus Holzer[FIX] AMD UsageSome minor fixes for using AMD GPUsSome minor fixes for using AMD GPUsMarkus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/622Add mpi-ext.h for MPIX_CUDA_AWARE_SUPPORT2023-07-26T09:51:23+02:00Samuel KemmlerAdd mpi-ext.h for MPIX_CUDA_AWARE_SUPPORTMarkus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/621remove msvc from CI2023-07-11T17:47:23+02:00Dominik Thoennesdominik.thoennes@fau.deremove msvc from CIremove msvc from CI since Windows images are currently broken
We might reintroduce the msvc in a docker imageremove msvc from CI since Windows images are currently broken
We might reintroduce the msvc in a docker imagehttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/620[Fix] MultiGPU Refinement2023-07-13T11:58:38+02:00Markus Holzer[Fix] MultiGPU RefinementMulti GPU mesh refinement did not work due to wrong access of the bufferMulti GPU mesh refinement did not work due to wrong access of the bufferMarkus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/619Check for cuda aware mpi2023-06-22T12:47:52+02:00Samuel KemmlerCheck for cuda aware mpiMarkus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/618Add weights to storage spec2023-06-27T10:23:13+02:00Markus HolzerAdd weights to storage specMarkus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/617Replace PyCuda with CuPy2023-08-02T11:24:36+02:00Markus HolzerReplace PyCuda with CuPyReplaces [PyCuda](https://documen.tician.de/pycuda/) with [CuPy](https://cupy.dev/)
Advantages of [CuPy](https://cupy.dev/):
- AMD support
- probably higher maintained due to NVIDIA support
- SciPy compatible.Replaces [PyCuda](https://documen.tician.de/pycuda/) with [CuPy](https://cupy.dev/)
Advantages of [CuPy](https://cupy.dev/):
- AMD support
- probably higher maintained due to NVIDIA support
- SciPy compatible.Markus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/616Changed the default CMAKE loglevel to INFO2023-06-16T11:41:26+02:00Christoph AltChanged the default CMAKE loglevel to INFOThis should fix #211This should fix #211https://i10git.cs.fau.de/walberla/walberla/-/merge_requests/615SVE-vector-length-bugfix2023-06-07T15:17:20+02:00Piet JarmatzSVE-vector-length-bugfixCompilation of waLBerla fails on our A64FX ARM systems, this merge request fixes it. It solves two problems:
1. SVE_LENGTH contains a newline character, this leads to a line break in cmake generated makefiles, thus to a "missing separa...Compilation of waLBerla fails on our A64FX ARM systems, this merge request fixes it. It solves two problems:
1. SVE_LENGTH contains a newline character, this leads to a line break in cmake generated makefiles, thus to a "missing separator" syntax error, so that try_compile fails. I used the REGEX REPLACE command to remove the line break, since the STRIP command seems to work only for some versions of CMake.
2. sve_default_vector_length returns a vector length in bytes, not in bits. So it has to be converted before it can be passed to msve-vector-bits, otherwise it leads to compiler errors.Markus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/614Fix compilation with gcc 132023-05-25T20:32:05+02:00Daniel BauerFix compilation with gcc 13A missing include makes compilation with gcc 13 fail.
This MR adds the missing include and reformats the affected file.A missing include makes compilation with gcc 13 fail.
This MR adds the missing include and reformats the affected file.Daniel BauerDaniel Bauerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/613Use force density in antidunes showcase2023-05-26T12:55:01+02:00Christoph SchwarzmeierUse force density in antidunes showcaseThis merge request adds
1. a corrected force density computation, which was assumed constant before.
2. the possibility to use different ways of distributing excessive mass in the FSLBM.This merge request adds
1. a corrected force density computation, which was assumed constant before.
2. the possibility to use different ways of distributing excessive mass in the FSLBM.Samuel KemmlerSamuel Kemmlerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/612set the loglevel to detail in ci2023-06-14T15:16:12+02:00Christoph Altset the loglevel to detail in ciThis MR sets the loglevel to detail in the CI as it was suggested in !609.
Fixes #178This MR sets the loglevel to detail in the CI as it was suggested in !609.
Fixes #178https://i10git.cs.fau.de/walberla/walberla/-/merge_requests/611[Fix] Initialise FlagField with one flag exclusively2023-05-09T15:46:46+02:00Philipp Suffa[Fix] Initialise FlagField with one flag exclusivelyFixing the problem of #203, so that the boundary initializer sets only one boundary flag per cell for custom boundary flags.
To set custom boundaries, use the forceBoundary function, if only one flag per cell is desired in the flag field.Fixing the problem of #203, so that the boundary initializer sets only one boundary flag per cell for custom boundary flags.
To set custom boundaries, use the forceBoundary function, if only one flag per cell is desired in the flag field.Philipp SuffaPhilipp Suffahttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/609Fix compiler errors in logging code2023-05-11T13:17:27+02:00Jean-Noël Gradjgrad@icp.uni-stuttgart.deFix compiler errors in logging codeDescription of changes:
* fix bugs in the logging mechanism revealed by !604
* several logging macros were missing a semicolon
* several calls to `WALBERLA_LOG_PROGRESS` were broken
* fix a few Clang-Tidy diagnosticsDescription of changes:
* fix bugs in the logging mechanism revealed by !604
* several logging macros were missing a semicolon
* several calls to `WALBERLA_LOG_PROGRESS` were broken
* fix a few Clang-Tidy diagnosticshttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/608Export Device selection to python2023-05-04T12:45:45+02:00Markus HolzerExport Device selection to pythonTo run python programs on multi-GPU with the python bindings the device selection must be exported.To run python programs on multi-GPU with the python bindings the device selection must be exported.Markus HolzerMarkus Holzerhttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/607Fix boundaries code generation2023-05-04T14:55:14+02:00Jean-Noël Gradjgrad@icp.uni-stuttgart.deFix boundaries code generationBoundary code generation now uses the correct floating point type and no longer declares unused variables.Boundary code generation now uses the correct floating point type and no longer declares unused variables.https://i10git.cs.fau.de/walberla/walberla/-/merge_requests/606Coupling tutorial2023-05-26T15:21:13+02:00Samuel KemmlerCoupling tutorialPhilipp SuffaPhilipp Suffahttps://i10git.cs.fau.de/walberla/walberla/-/merge_requests/605Resolve "Remove remaining boost components"2023-06-14T15:17:44+02:00Dominik Thoennesdominik.thoennes@fau.deResolve "Remove remaining boost components"Closes #190Closes #190https://i10git.cs.fau.de/walberla/walberla/-/merge_requests/604document logging2023-06-14T15:35:47+02:00Christoph Altdocument loggingThis is my suggestion for #178.
This set the init value of the runtime log level to the same value as it is set in CMake.
Further, the warning in the setLogLevel is a bit more explicit.
I did not changed the default log level, as I foun...This is my suggestion for #178.
This set the init value of the runtime log level to the same value as it is set in CMake.
Further, the warning in the setLogLevel is a bit more explicit.
I did not changed the default log level, as I found this solution more intuitive to have the to loglevels (initialy) in-sync.https://i10git.cs.fau.de/walberla/walberla/-/merge_requests/603Resolve "Only using WALBERLA_BUILD_WITH_PARMETIS=ON fails"2023-05-11T09:45:22+02:00Dominik Thoennesdominik.thoennes@fau.deResolve "Only using WALBERLA_BUILD_WITH_PARMETIS=ON fails"Closes #195Closes #195Christoph AltChristoph Alt