Expand description
Ruda parallel primitives.
Modules§
- block
- Block collectives. Every thread in the block participates in every call.
- collective
- Operators shared by device, block and logical-warp collectives.
- reduce
- This provides different implementations of the reduce algorithm which can run on multiple GPU backends using Ruda.
- warp
- Collectives over logical warps within a native device subgroup.
Macros§
- ruda_
decomposer - Define a lexicographic arithmetic-field decomposition of any width. Fields are listed most significant first; padding is never a sort key.
- ruda_
record - Define a C-layout record, including nested records, for primitive operators.