Class ExecutionContext

• Defined in File execution_context.hpp

Derived Type

• public holoscan::gxf::GXFExecutionContext (Class GXFExecutionContext)

class ExecutionContext

Class to hold the execution context.

This class provides the execution context for the operator.

Subclassed by holoscan::gxf::GXFExecutionContext

Public Functions

ExecutionContext() = default

Construct a new Execution Context object.

virtual ~ExecutionContext() = default

inline std::shared_ptr<InputContext> input() const

Get the input context.

Returns

The shared pointer to the input context.

inline std::shared_ptr<OutputContext> output() const

Get the output context.

Returns

The shared pointer to the output context.

inline void *context() const

Get the context.

Returns

The pointer to the context.

inline virtual expected<cudaStream_t, RuntimeError> allocate_cuda_stream(const std::string &stream_name = "")

Allocate a CUDA stream from the operator’s CudaStreamPool.

Streams are cached by name — calling with the same name returns the same stream on subsequent calls within the same operator. This is useful for root operators that need to allocate a stream (rather than receiving one from upstream).

Streams allocated this way are not automatically emitted on output ports. Call OutputContext::set_cuda_stream() before emit() if you need to propagate the stream.

Parameters

stream_name – A name for the stream. The same name returns the same stream on subsequent calls. Defaults to an empty string.

Returns

The allocated cudaStream_t, or an error if no CudaStreamPool is available.

inline virtual void synchronize_streams(const std::vector<std::optional<cudaStream_t>> &cuda_streams, cudaStream_t target_cuda_stream)

Synchronize multiple CUDA streams to a target stream (non-blocking).

Uses cudaEventRecord and cudaStreamWaitEvent to create GPU-side dependencies without blocking the CPU. This is the same mechanism used internally by receive_cuda_stream.

When using receive_cuda_stream, synchronization is handled automatically and this method is not needed. It is provided for advanced manual stream handling use cases.

Parameters

• cuda_streams – Vector of streams to synchronize. std::nullopt elements are skipped.

• target_cuda_stream – The stream that will wait for all other streams to complete.

inline virtual expected<int, RuntimeError> device_from_stream(cudaStream_t stream)

Get the CUDA device ID for a given stream.

Only works with Holoscan-managed streams (those returned by receive_cuda_stream, receive_cuda_streams, or allocate_cuda_stream).

Parameters

stream – The CUDA stream to query.

Returns

The device ID, or an error if the stream is not managed by Holoscan.

inline virtual bool is_gpu_available() const

check if GPU capability is present on the system

inline virtual std::shared_ptr<Operator> find_operator(const std::string &op_name = "")

Find an operator by name.

If the operator name is not provided, the current operator is returned.

Parameters

op_name – The name of the operator.

Returns

A shared pointer to the operator or nullptr if the operator is not found.

inline virtual expected<holoscan::OperatorStatus, RuntimeError> get_operator_status(const std::string &op_name = "")

Get the status of the operator.

If the operator name is not provided, the status of the current operator is returned.

Parameters

op_name – The name of the operator.

Returns

The status of the operator or an error if the operator is not found.

Protected Attributes

std::shared_ptr<InputContext> input_context_ = nullptr

The input context.

std::shared_ptr<OutputContext> output_context_ = nullptr

The output context.

void *context_ = nullptr

The context.