NVIDIA Docs Hub Homepage NVIDIA Holoscan Holoscan SDK v4.0.0 Function holoscan::inference::build_execution_plan

Function holoscan::inference::build_execution_plan

Defined in File holoinfer_utils.hpp

Function Documentation

InferStatus holoscan::inference::build_execution_plan(const MultiMappings &pre_processor_map, const MultiMappings &inference_map, std::vector<std::vector<std::string>> &execution_plan)

Build execution plan.

Parameters

pre_processor_map – Map with model name as key, mapped to vector of input tensor names
inference_map – Map with model name as key, mapped to vector of output tensor names
execution_plan – Vector of levels; each level contains models that can run in parallel.

Returns

InferStatus with appropriate code and message

Previous Function holoscan::inference::build_engine

Next Function holoscan::inference::check_cuda