uni20::async::ICudaScheduler Class Reference

Scheduler interface for initial CudaTask admission. More...

#include <cuda_task.hpp>

Inheritance diagram for uni20::async::ICudaScheduler:

Collaboration diagram for uni20::async::ICudaScheduler:

Public Member Functions
virtual void	schedule (CudaTask &&task)=0
	Submit a CUDA task without establishing device affinity.
virtual void	schedule (CudaTask &&task, int device)=0
	Bind and submit a CUDA task for initial execution on a device.
	~ICudaScheduler () override=default
Public Member Functions inherited from uni20::async::IScheduler
virtual	~IScheduler ()=default
	Virtual destructor.

Detailed Description

Scheduler interface for initial CudaTask admission.

Virtual inheritance permits one scheduler object to implement both ordinary and CUDA admission with one unambiguous internal route.

Constructor & Destructor Documentation

~ICudaScheduler()

uni20::async::ICudaScheduler::~ICudaScheduler ( )

overridedefault

Member Function Documentation

schedule() [1/2]

virtual void uni20::async::ICudaScheduler::schedule ( CudaTask &&  task )

pure virtual

Submit a CUDA task without establishing device affinity.

The scheduler routes the task through its default CUDA device until the task explicitly selects a device.

Parameters

task	CUDA task to admit.

Implemented in uni20::async::DebugCudaScheduler, and uni20::async::TbbCudaScheduler.

schedule() [2/2]

virtual void uni20::async::ICudaScheduler::schedule ( CudaTask &&  task, int  device  )

pure virtual

Bind and submit a CUDA task for initial execution on a device.

Parameters

task	CUDA task to admit.
device	CUDA runtime device ordinal.

Implemented in uni20::async::DebugCudaScheduler, and uni20::async::TbbCudaScheduler.

---

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

• cuda_task.hpp