VTK-m  2.1
Public Member Functions | Private Member Functions | Private Attributes | List of all members
vtkm::cont::DeviceAdapterTimerImplementation< vtkm::cont::DeviceAdapterTagCuda > Class Reference

Specialization of DeviceAdapterTimerImplementation for CUDA CUDA contains its own high resolution timer that are able to track how long it takes to execute async kernels. More...

#include <DeviceAdapterTimerImplementationCuda.h>

Public Member Functions

 DeviceAdapterTimerImplementation ()
 
 ~DeviceAdapterTimerImplementation ()
 
void Reset ()
 
void Start ()
 
void Stop ()
 
bool Started () const
 
bool Stopped () const
 
bool Ready () const
 
vtkm::Float64 GetElapsedTime () const
 

Private Member Functions

 DeviceAdapterTimerImplementation (const DeviceAdapterTimerImplementation< vtkm::cont::DeviceAdapterTagCuda > &)=delete
 
void operator= (const DeviceAdapterTimerImplementation< vtkm::cont::DeviceAdapterTagCuda > &)=delete
 

Private Attributes

bool StartReady
 
bool StopReady
 
cudaEvent_t StartEvent
 
cudaEvent_t StopEvent
 

Detailed Description

Specialization of DeviceAdapterTimerImplementation for CUDA CUDA contains its own high resolution timer that are able to track how long it takes to execute async kernels.

If we simply measured time on the CPU it would incorrectly just capture how long it takes to launch a kernel.

Constructor & Destructor Documentation

◆ DeviceAdapterTimerImplementation() [1/2]

◆ ~DeviceAdapterTimerImplementation()

◆ DeviceAdapterTimerImplementation() [2/2]

Member Function Documentation

◆ GetElapsedTime()

◆ operator=()

◆ Ready()

◆ Reset()

◆ Start()

◆ Started()

◆ Stop()

◆ Stopped()

Member Data Documentation

◆ StartEvent

◆ StartReady

◆ StopEvent

◆ StopReady


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