Struct CudaBufferHandle 

pub struct CudaBufferHandle { /* private fields */ }

1D (flat) CUDA device pointer with byte size.

Implementations

impl CudaBufferHandle

pub unsafe fn try_from_raw( ptr: u64, size: u64, uuid: Option<[u8; 16]>, keep_alive: Option<KeepAlive>, ) -> Result<Self, InvalidHandleError>

Wrap a flat CUdeviceptr with an explicit byte size. Borrowing: the handle is stored non-owningly, this type has no Drop, and every obligation below extends to each Clone.

Safety

• ptr must be a live CUdeviceptr addressing at least size contiguous bytes of currently-allocated device memory. Consumers read and write the whole [ptr, ptr + size) range, so an over-stated size is an out-of-bounds device access.
• ptr == 0 is rejected as InvalidHandleError::NullHandle and size == 0 as InvalidHandleError::ZeroSize; neither is a caller obligation.
• Context affinity, uuid, lifetime (no retain, no Drop, must outlive every Clone, discharged by keep_alive) and thread rules are exactly those of CudaPtr2D::try_from_raw.

pub fn ptr(&self) -> u64

pub fn size(&self) -> u64

pub fn uuid(&self) -> Option<[u8; 16]>

pub fn keep_alive(&self) -> Option<&KeepAlive>

Trait Implementations

impl Clone for CudaBufferHandle

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CudaBufferHandle

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

Formats the value using the given formatter.

impl Send for CudaBufferHandle

impl Sync for CudaBufferHandle