Struct ExecutionContext 

pub struct ExecutionContext<'e> { /* private fields */ }

Implementations

impl ExecutionContext<'_>

pub fn as_raw(&self) -> trtIExecutionContext_t

pub fn set_input_shape(&self, name: &str, dims: Dims) -> Result<()>

pub unsafe fn set_tensor_address( &self, name: &str, addr: *mut c_void, ) -> Result<()>

Bind a device pointer to a named input/output tensor on this execution context. The pointer is forwarded to TensorRT’s setTensorAddress, which uses it during enqueueV3 execution.

Safety

addr must point to device memory that:

• is large enough for the tensor’s bound shape and data type, and
• remains valid (not freed, not unmapped) for the duration of any enqueueV3 call that runs after this binding and before the stream completes.

pub fn tensor_shape(&self, name: &str) -> Result<Dims>

pub fn tensor_address(&self, name: &str) -> Result<*mut c_void>

Read the current bound device address for a tensor (null if unset).

pub unsafe fn enqueue_v3(&self, stream: cudaStream_t) -> Result<()>

Enqueue the inference on the given CUDA stream. Returns Ok if TRT reports success; the stream is still responsible for ordering, and the caller must ensure all tensor addresses have been set.

Safety

stream must be a valid cudaStream_t that outlives the enqueue.

Trait Implementations

impl<'e> Debug for ExecutionContext<'e>

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

Formats the value using the given formatter.

impl Drop for ExecutionContext<'_>

fn drop(&mut self)

Executes the destructor for this type.

fn pin_drop(self: Pin<&mut Self>)

🔬This is a nightly-only experimental API. (pin_ergonomics)

Execute the destructor for this type, but different to Drop::drop, it requires self to be pinned.