| | 1 | == Research Questions == |
| | 2 | The basic idea of this study is to explore questions like the following: |
| | 3 | - How does defect proneness vary with concurrency dialect? |
| | 4 | -- number of defects |
| | 5 | -- type of defect |
| | 6 | -- difficulty of localizing defect and other measures |
| | 7 | - Does verification cost and effectiveness vary with concurrency dialect? |
| | 8 | -- time/space/... |
| | 9 | -- precision of results |
| | 10 | |
| | 11 | |
| | 12 | == Sample Systems to Study == |
| | 13 | - https://github.com/UK-MAC/CloverLeaf |
| | 14 | -- CUDA, MPI, OpenMP, OpenACC, MPI+OpenMP |
| | 15 | -- Serial version for comparison |
| | 16 | -- version history to study errors over time |
| | 17 | - https://github.com/HydroBench/Hydro/tree/master/HydroC |
| | 18 | -- OpenMP+MPI, OpenACC+MPI, CUDA+MPI, OpenACC+MPI, OpenCL+MPI |
| | 19 | - http://op2.github.io/PyOP2/backends.html |
| | 20 | -- OpenMP, CUDA, OpenCL, MPI, MPI+OpenMP |
| | 21 | -- not a standalone app, but rather a backend for computation |
| | 22 | |
