Skip to main content

map_array_async

baracuda_driver::memcpy3d

Function map_array_async 

Source 
pub fn map_array_async(
    info: &mut [CUarrayMapInfo],
    stream: &Stream,
) -> Result<()>
Expand description

Bulk map / unmap sparse-array tiles and mipmap tails on a stream.

Each CUarrayMapInfo entry describes one tile update — use the builder methods on that type (with_array / with_mipmapped_array, with_sparse_level / with_miptail, with_mem_handle, as_map / as_unmap, with_offset) to construct entries, then pass the array here. Mix map and unmap ops in a single call freely.

Availability: sparse-array support requires a device with sparse residency (compute capability 7.0+ on desktop; Ampere+ for full mipmap-tail support). Calls on non-supporting devices fail with CUDA_ERROR_NOT_SUPPORTED.