Closed lucifer1004 closed 9 months ago
This PR implements a warp-level (sub-warp is also supported via the template parameter TileSize, but there is no perf gain) parallel algorithm for the Monte Carlo kernel.
TileSize
This PR implements a warp-level (sub-warp is also supported via the template parameter
TileSize
, but there is no perf gain) parallel algorithm for the Monte Carlo kernel.