Struct Kernel 

pub struct Kernel { /* private fields */ }

A kernel entry point inside a Library.

Implementations

impl Kernel

pub fn launch(&self) -> LaunchBuilder<'_>

Start a kernel-launch builder for this kernel.

impl Kernel

pub fn as_raw(&self) -> cudaKernel_t

Raw cudaKernel_t. Use with care.

pub fn as_launch_ptr(&self) -> *const c_void

Returns the raw kernel handle cast to a const void* — the form expected by cudaLaunchKernel. Library-loaded kernels can be launched through the standard runtime launch function by passing this pointer.

pub fn max_active_blocks_per_multiprocessor( &self, block_size: i32, dynamic_smem_bytes: usize, ) -> Result<i32>

cudaOccupancyMaxActiveBlocksPerMultiprocessor — how many blocks of size block_size can run concurrently per SM given dynamic_smem_bytes of dynamic shared memory.

pub fn max_active_blocks_per_multiprocessor_with_flags( &self, block_size: i32, dynamic_smem_bytes: usize, flags: u32, ) -> Result<i32>

cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags — same as Self::max_active_blocks_per_multiprocessor but accepting occupancy-flag bits (0 = default, 1 = disable shared-memory carveout adjustment).

pub fn available_dynamic_smem_per_block( &self, num_blocks: i32, block_size: i32, ) -> Result<usize>

cudaOccupancyAvailableDynamicSMemPerBlock — how much dynamic shared memory can each block use if num_blocks of block_size threads run concurrently on each SM.

pub fn set_attribute(&self, attr: i32, value: i32) -> Result<()>

Set a writable cudaFuncAttribute. The common one is cudaFuncAttributeMaxDynamicSharedMemorySize = 8.

Trait Implementations

impl Clone for Kernel

fn clone(&self) -> Kernel

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Kernel

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Send for Kernel

impl Sync for Kernel