Releases: SeisSol/SeisSol
Releases · SeisSol/SeisSol
v1.1.3
Summary
- MPI using only GPU buffers
- Instantaneous Time Mirror support
- Many bugfixes
What's Changed
- improved interpolation for FL33 by @Thomas-Ulrich in #734
- Update recreateXdmf.py by @vikaskurapati in #994
- detail the dumping of dynamic libraries by @Thomas-Ulrich in #969
- Add gfx942 to the CMake files by @davschneller in #998
- Disable Some OpenMP Clauses for NVHPC by @davschneller in #999
- Implementation of Instantaneous time mirror for SeisSol by @vikaskurapati in #995
- [GPU] Move buffersDerivatives bucket to device memory instead of shared memory by @uphoffc in #982
- Persistent MPI by @davschneller in #997
- Fix GPU fault output by @davschneller in #1001
- Update the Performance Metric Documentation by @davschneller in #1002
- Vikas/parameter doc by @vikaskurapati in #1005
- Refactor the Loop Statistics Implementation by @davschneller in #952
- Remove superfluous include by @davschneller in #1007
- Remove Unnecessary Fault Output Allocations by @davschneller in #1011
- Remove duplicate related publication by @Galasnow in #1013
- Re-Enable CI by @davschneller in #1017
- Remove unnecessary header by @davschneller in #1018
- Small CMake Fixes by @davschneller in #1015
- Bump puml to latest version by @sebwolf-de in #1020
- Bump cpu image version by @sebwolf-de in #1022
- Add more timing infos by @davschneller in #1025
- Adjust fault face indexing to not iterate over an
std::map
for each lookup by @davschneller in #1027 - Fix wrong array indexing (introduced by #1027) by @davschneller in #1028
New Contributors
- @vikaskurapati made their first contribution in #994
Full Changelog: v1.1.2...v1.1.3
v1.1.2
Summary
Another bugfix release.
What's Changed
- Allow cubeGenerator to be used as a mesh generator in SeisSol by @montrie in #918
- fixes yield function in tpv13.rst by @Thomas-Ulrich in #981
- Fix bug in free surface integrator that occurs if SeisSol alignment != yateto alignment by @uphoffc in #983
- add isochrone script by @NicoSchlw in #896
- Adjust tensor sizes to fix plasticity error by @davschneller in #985
- Refactor source time function in kinematic rupture by @uphoffc in #976
- Always compile with comm thread and mini SeisSol enabled by @davschneller in #986
- Update .readthedocs.yml by @davschneller in #987
- Update CMakeLists.txt by @davschneller in #989
- fail if dynamic rupture parameters not find by @Thomas-Ulrich in #988
- change time step selection of extractDataFromUnstructuredOutput.py by @Thomas-Ulrich in #990
- refactor code calculating fault energies and fix associated documentation by @Thomas-Ulrich in #849
Full Changelog: v1.1.1...v1.1.2
v1.1.1
Summary
A bugfix release. The main "smaller" changes are:
- first official support for MacOS #963
- faster code generation by YATeTo, support for more CPU architectures #962
- several bugfixes found during large-scale runs #970 #972
Note that this version is required to support PSpaMM in its Python package variant (SeisSol/PSpaMM#8 ), since the name of the generator program needed to change to not conflict with the module name.
What's Changed
- Improve ASAGI cmake by @sebwolf-de in #959
- Update CPU Architectures, Update YATeTo by @davschneller in #962
- Add support for MacOS & M1/M2 by @krenzland in #963
- Improving fault output documentation by @AliceGabriel in #868
- Remove ThreadPinningWriter warning on non-Apple systems by @davschneller in #964
- Install SeisSol into bin and lib dirs by @davschneller in #967
- Only create FLOP counter file for rank==0 by @davschneller in #970
- Add (shorter) CI expiration time span for artifacts by @davschneller in #971
- Remove new sharedmem comm allocation for node mask by @davschneller in #972
- Update find libxsmm by @uphoffc in #975
- fix index in corner frequency script by @NicoSchlw in #968
- Propagate Libxsmm and PSpaMM path from CMake to the generator scripts / Yateto by @davschneller in #977
Full Changelog: v1.1.0...v1.1.1
v1.1.0
Main New Features And Changes
- Poroelastic media now supports dynamic rupture #895
- Internally, we removed the code that was still written in FORTRAN, now everything is written in C++ #829
- Point sources can now be computed on GPUs #876
A Detailed Log Of What's Changed
- Added a sycl-independent interface for gpu-dr by @ravil-mobile in #838
- fixing std::filesystem by @ravil-mobile in #843
- Fix oneapi build by @uphoffc in #845
- Added
SEISSOL_FREE_CPUS_MASK
env. variable by @ravil-mobile in #848 - Added additional cube boundary options by @montrie in #852
- Added an option to change locations of the MPI buffers for the communication layers by @ravil-mobile in #847
- Make
target_sources
in CMake private by @davschneller in #851 - Support for alternative graph partitioning libs by @uphoffc in #844
- Fix read the docs build by @uphoffc in #853
- Fixes required for lumi by @ravil-mobile in #855
- Fixed
0
byte allocation in sycl (dr) by @ravil-mobile in #856 - Replace the Remaining Parts of the FORTRAN Initialization Procedure by @davschneller in #829
- Cleanup CMakeLists by @davschneller in #861
- Set device context for the communication thread by @ravil-mobile in #860
- Updated the reference to Device by @ravil-mobile in #862
- Change slipRateOutputType to 0 when using a friction law with imposed slip rate by @Thomas-Ulrich in #858
- use seissol::MPI to avoid ambiguity by @uphoffc in #866
- Improve 'Pick components' filter by @sebwolf-de in #864
- Update slip-rate-on-DR.rst by @jwjeremy in #863
- A bit more descriptive Git version display by @davschneller in #859
- Fix some input parameters (incl. RFileName) by @davschneller in #871
- Fix Receiver Output Length Error by @davschneller in #875
- Update AUTHORS and CITATION files by @davschneller in #880
- Moved device initialization to the very beginning by @ravil-mobile in #879
- Adapted the device::UsmAllocator by @ravil-mobile in #881
- Point source GPU by @uphoffc in #876
- DR/GPU: Improved copy and compute energy only when enabled by @ravil-mobile in #885
- Update first example by @sebwolf-de in #887
- Updated the reference to Device by @ravil-mobile in #892
- Fix namespace by @uphoffc in #894
- make script compatible with volume output by @NicoSchlw in #826
- Updated reference to yateto and chainforge by @ravil-mobile in #898
- Re-order tv init and time cluster creation by @davschneller in #900
- Fix CMake for Not Being in a Git Repository by @davschneller in #877
- Bump scipy from 1.7.3 to 1.10.0 in /postprocessing/science/GroundMotionParametersMaps by @dependabot in #901
- Added
showEdgeCutStatistics
option by @ravil-mobile in #893 - Synchronize LTSTree after Setup of the Initial Conditions by @davschneller in #903
- Fix NRF reader test by @sebwolf-de in #904
- Updated docu and gitlab-ci pipelines by @ravil-mobile in #907
- Update filesystem.h by @ravil-mobile in #909
- Fix BoundingBox parameter parsing (and other arrays) by @davschneller in #912
- Prepare to Apply
clang-tidy
to More Source Files by @davschneller in #914 - Fix plasticity kernel. Added missing
updateRelaxTime()
by @ravil-mobile in #916 - Fix
ZinvData
being used after going out of scope by @davschneller in #917 - Remove FaultMPI by @davschneller in #915
- Add A64FX support by @krenzland in #919
- Remove
FaultOutputFlag
andBC_dr
Input Parameters by @davschneller in #924 - Optionally Serialize Device Initialization by @davschneller in #923
- Fix init travelling wave by @krenzland in #927
- GPU Flux Premultiplication by @davschneller in #921
- document cuda installation workflow for heisenbug by @Thomas-Ulrich in #931
- Update Gitlab pipelines by @davschneller in #934
- Adjust CI Pipelines (again) by @davschneller in #935
- Fix typo in documentation by @Galasnow in #936
- Fix 940 by @sebwolf-de in #941
- dr/poro by @sebwolf-de in #895
- Adding
fixtimestep
Restriction for Viscoelastic by @davschneller in #942 - More CI benchmarks by @sebwolf-de in #943
- Change the doxygen build CI instance by @davschneller in #945
- Make SYCL choosable in CMake, and make NVHPC choosable for HIPSYCL/Open SYCL + Nvidia GPUs by @davschneller in #922
- Same time stamp in backup files/output folder creation by @ravil-mobile in #884
- Backup XDMF Files by @sebwolf-de in #950
- Fix A64FX alignment by @krenzland in #953
- Allow retry for GPU build by @sebwolf-de in #951
- Replace
NUMBER_OF_QUANTITIES
by the correct dimension by @davschneller in #958
New Contributors
- @montrie made their first contribution in #852
- @davschneller made their first contribution in #851
- @jwjeremy made their first contribution in #863
- @Galasnow made their first contribution in #936
Full Changelog: v1.0.1...v1.1.0
v1.0.1
What's Changed
- fixed docu by @ravil-mobile in #834
- Removed deprecated
10core-docker-runner
runner from the pipeline by @ravil-mobile in #836 - add get_distances_to_rupture.jl by @NicoSchlw in #811
- Fix off by one error in cluster merging by @krenzland in #837
Full Changelog: v1.0.0...v1.0.1
v1.0.0
SeisSol Release v1.0.0. Major new features since v0.9.0:
- Dynamic Rupture routines in C++ (including GPU offloading with SYCL) #663
- Actor implementation of LTS with wiggle factor option #758
- Include check for mesh correctness #781
- remove positive state enforcement #752
- Material homogenization by cell-averaging #644
- Implemented GravitationalFreeSurfaceBC for GPUs #749
- Compute rotation for receiver output #748
- Improved profiling #821
- Add CITATION.cff #827
- A lot of bug fixes
- A lot of new documentation
v1.0.0-rc2
Second release candidate for v1.0.0. Major new features:
- Dynamic Rupture routines in C++ (including GPU offloading with SYCL) #663
- Actor implementation of LTS with wiggle factor option #758
- Include check for mesh correctness #781
- remove positive state enforcement #752
- Material homogenization by cell-averaging #644
- Implemented
GravitationalFreeSurfaceBC
for GPUs #749 - Compute rotation for receiver output #748
- A lot of bug fixes
- A lot of new documentation
v1.0.0-rc
The release candidate of the C++ implementation of Dynamic Rupture
- contains OpenMP offloading implementation of Dynamic Rupture
v0.9.0
This is the last snapshot of SeisSol which contains the Dynamic Rupture implementation written in Fortran
SC21 Revision
sc21revision Use different approach to skip mini seissol