cuda::experimental::stf::tiled

template<size_t tile_size, typename mdspan_shape_t>
auto cuda::experimental::stf::tiled(

const mdspan_shape_t s,

size_t part_id,

size_t nparts,

)

Tiled partition strategy applied on a shape of mdspan.

Since there is no partial template deduction on classes, we provide a function to implement tiled<tile_size> and have the other type deduced.

Template Parameters:

• `tile_size` – size of the tiles

• `mdspan_shape_t` – shape of a mdspan