tf::cudaSequentialCapturing Class Reference

class to capture a CUDA graph using a sequential stream More...

#include <cuda_optimizer.hpp>

Inheritance diagram for tf::cudaSequentialCapturing:

Collaboration diagram for tf::cudaSequentialCapturing:

Public Member Functions
	cudaSequentialCapturing ()=default
	constructs a sequential optimizer

Friends
class	cudaFlowCapturer

Detailed Description

class to capture a CUDA graph using a sequential stream

A sequential capturing algorithm finds a topological order of the described graph and captures dependent GPU tasks using a single stream. All GPU tasks run sequentially without breaking inter dependencies.

---

The documentation for this class was generated from the following file:

• cuda_optimizer.hpp