Data Parallelism - ajpowelsnl/kokkos GitHub Wiki
Parallel execution patterns for composing algorithms.
| Function | Description |
|---|---|
| parallel_for | Executes user code in parallel |
| parallel_reduce | Executes user code to perform a reduction in parallel |
| parallel_scan | Executes user code to generate a prefix sum in parallel |
| fence | Fences execution spaces |
Reducer objects used in conjunction with parallel_reduce.
| Reducer | Description |
|---|---|
| BAnd | Binary 'And' reduction |
| BOr | Binary 'Or' reduction |
| LAnd | Logical 'And' reduction |
| LOr | Logical 'Or' reduction |
| Max | Maximum reduction |
| MaxLoc | Reduction providing maximum and an associated index |
| Min | Minimum reduction |
| MinLoc | Reduction providing minimum and an associated index |
| MinMax | Reduction providing both minimum and maximum |
| MinMaxLoc | Reduction providing both minimum and maximum and associated indices |
| Prod | Multiplicative reduction |
| Sum | Sum reduction |
Tags passed to team_size_max or team_size_recommended to identify whether the pattern for which the team size for a functor is being requested
| Tag | pattern |
|---|---|
| ParallelForTag | parallel_for |
| ParallelReduceTag | parallel_reduce |
| ParallelScanTag | parallel_scan |