Instrumentor Class Reference

Basic instrumentation profiler for Google Chrome tracing format. More...

#include <Instrumentor.h>

Public Member Functions
	Instrumentor ()
void	beginSession (const std::string &name, const bool storeInMemory=false, const std::string &filePath="Profiling-Results.json")
void	endSession ()
void	addProfile (const ProfileResult &result)
void	writeProfile (const ProfileResult &result)
void	writeHeader ()
void	writeFooter ()
std::string	filePath ()

Static Public Member Functions
static Instrumentor &	get ()

Private Attributes
InstrumentationSession *	_currentSession
std::string	_filePath
std::ofstream	_outputStream
int	_profileCount
std::mutex	_mutex
bool	_storeInMemory = false
std::vector< ProfileResult >	_profileResults

Detailed Description

Basic instrumentation profiler for Google Chrome tracing format.

Usage: include this header file somewhere in your code. In your most outer function (e.g. main) you have to begin profiling session with: Instrumentor::get().beginSession("Session Name"); If you pass storeInMemory=true the profileResults will be stored in memory instead of being written into the file stream which is pretty slow. Of course the in memory storage can quickly use a lot of memory depending how fine grained your profiling is. In app-Demo-SLProject this is done in SLInterface::slCreateAppAndScene.

In between you can add either PROFILE_FUNCTION(); at the beginning of any routine or PROFILE_SCOPE(scopeName) at the beginning of any scope you want to measure.

At the end of your most outer function (e.g. main) you end the session with: Instrumentor::get().endSession();

After the endSession you can drag the Profiling-Results.json file into the chrome://tracing page of the Google Chrome browser. In app-Demo-SLProject this is done in SLInterface::slTerminate.

Constructor & Destructor Documentation

Instrumentor()

Instrumentor::Instrumentor ( )

inline

Member Function Documentation

addProfile()

void Instrumentor::addProfile ( const ProfileResult &  result )

inline

addProfile should be as fast as possible for not influencing the profiling by the profiler itself. In addition it must be thread safe.

beginSession()

void Instrumentor::beginSession ( const std::string &  name, const bool  storeInMemory = false, const std::string &  filePath = "Profiling-Results.json"  )

inline

endSession()

void Instrumentor::endSession ( )

inline

filePath()

std::string Instrumentor::filePath ( )

inline

get()

static Instrumentor & Instrumentor::get ( )

inlinestatic

writeFooter()

void Instrumentor::writeFooter ( )

inline

writeHeader()

void Instrumentor::writeHeader ( )

inline

writeProfile()

void Instrumentor::writeProfile ( const ProfileResult &  result )

inline

Member Data Documentation

_currentSession

InstrumentationSession* Instrumentor::_currentSession

private

_filePath

std::string Instrumentor::_filePath

private

_mutex

std::mutex Instrumentor::_mutex

private

_outputStream

std::ofstream Instrumentor::_outputStream

private

_profileCount

int Instrumentor::_profileCount

private

_profileResults

std::vector<ProfileResult> Instrumentor::_profileResults

private

_storeInMemory

bool Instrumentor::_storeInMemory = false

private

---

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

• Instrumentor.h