- Dynamic load balancing
Dynamic load balancing
- Linearise and fuse GPU data
Linearise and fuse GPU data
- Real-time tasking and tasking priorities in general
Real-time tasking and tasking priorities in general
- cleanup done
- quite some cleanup required
- fixed typo
- Merge branch 'ader-dg' into 'p4'
Visualisation supports ADER-DG (and DG) now See merge request !225
- Made vis handle ader-dg
- Start to rewrite the conversion scripts in Python
- Merge branch 'ccz4-parameters' into 'p4'
Ccz4 parameters See merge request !223
- Merge branch 'makebedehappy' into 'p4'
Makebedehappy See merge request !224
- On Bede, llvm errors on Vector<Dimension,double> assignment so I replace the…
On Bede, llvm errors on Vector assignment so I replace the function calls with explicit assignments and function bodies dumped into the code. For posterity, the error is with regards to : 'variab …
All continuous integration data below stems from Peano’s fourth generation (P4). The stable, frozen third generation is not covered by CI.
Peano4 requires MPI_THREAD_MULTIPLE in the MPI implementation. If a server does not provide this level, the unit tests may pass (otherwise we display a fail), but the code won't be as efficient as it could be.
Performance regression tests
Regular grid on the unit square. No simulation data/PDE solved but the code adds an artificial number of floating point operations per cell.