ModuleBase::timer Class Reference

Tracing computation time. More...

#include <timer.h>

Collaboration diagram for ModuleBase::timer:

Classes
struct	Timer_One

Static Public Member Functions
static void	start (const std::string &class_name_in, const std::string &name_in)
	Use twice at a time: the first time, set start_flag to false; the second time, calculate the time duration.
static void	end (const std::string &class_name_in, const std::string &name_in)
static void	start (void)
	Start total time calculation.
static void	finish (std::ofstream &ofs, const bool print_flag=true, const bool check_end=true)
	Finish total time calculation and print computational processes with duration > 0.1 s.
static void	enable (void)
	Enable time computation.
static void	set_nvtx_enabled (bool b)
	Toggle NVTX range emission for CUDA profiling. Caller-injected; only consulted when built with __CUDA && __USE_NVTX.
static void	disable (void)
	Disable time computation.
static void	write_to_json (std::string file_name)
	Write all computational processes to json file.
static void	print_all (std::ofstream &ofs, const bool check_end)
	Print all computational processes with during > 0.1 s.
static long double	print_until_now (void)
	Stop total time calculation, print total time until now, and then start total time calculation again.

Static Public Attributes
static std::map< std::string, std::map< std::string, Timer_One > >	timer_pool

Static Private Member Functions
static double	cpu_time (void)
	Member function: calculate time.

Static Private Attributes
static bool	disabled = false
	Member variable: if disabled , timer can't work.
static bool	enable_nvtx_ = false
	Member variable: NVTX range emission toggle (CUDA profiling only).
static size_t	n_now = 0
	Member variable: the index of clocks.

Detailed Description

Tracing computation time.

Authors

Fangwei, Mohan, Peize Lin

Member Function Documentation

cpu_time()

double ModuleBase::timer::cpu_time ( void  )

staticprivate

Member function: calculate time.

Returns

double

Here is the caller graph for this function:

disable()

static void ModuleBase::timer::disable ( void  )

inlinestatic

Disable time computation.

Here is the caller graph for this function:

enable()

static void ModuleBase::timer::enable ( void  )

inlinestatic

Enable time computation.

end()

void ModuleBase::timer::end ( const std::string &  class_name_in, const std::string &  name_in  )

static

Here is the call graph for this function:

finish()

void ModuleBase::timer::finish ( std::ofstream &  ofs, const bool  print_flag = true, const bool  check_end = true  )

static

Finish total time calculation and print computational processes with duration > 0.1 s.

Parameters

ofs	The output file for print out timings
print_flag	Print timings or not

Here is the call graph for this function:

Here is the caller graph for this function:

print_all()

void ModuleBase::timer::print_all ( std::ofstream &  ofs, const bool  check_end  )

static

Print all computational processes with during > 0.1 s.

Parameters

ofs	The output file for print out timings

Here is the call graph for this function:

Here is the caller graph for this function:

print_until_now()

long double ModuleBase::timer::print_until_now ( void  )

static

Stop total time calculation, print total time until now, and then start total time calculation again.

Returns

long double

Here is the call graph for this function:

Here is the caller graph for this function:

set_nvtx_enabled()

static void ModuleBase::timer::set_nvtx_enabled ( bool  b )

inlinestatic

Toggle NVTX range emission for CUDA profiling. Caller-injected; only consulted when built with __CUDA && __USE_NVTX.

Here is the caller graph for this function:

start() [1/2]

void ModuleBase::timer::start ( const std::string &  class_name_in, const std::string &  name_in  )

static

Use twice at a time: the first time, set start_flag to false; the second time, calculate the time duration.

Parameters

class_name_in	The class name for timing
name_in	The compuational process for timing

Here is the call graph for this function:

start() [2/2]

void ModuleBase::timer::start ( void  )

static

Start total time calculation.

Here is the call graph for this function:

write_to_json()

void ModuleBase::timer::write_to_json ( std::string  file_name )

static

Write all computational processes to json file.

Parameters

file_name

The output file name

Here is the call graph for this function:

Here is the caller graph for this function:

Member Data Documentation

disabled

bool ModuleBase::timer::disabled = false

staticprivate

Member variable: if disabled , timer can't work.

enable_nvtx_

bool ModuleBase::timer::enable_nvtx_ = false

staticprivate

Member variable: NVTX range emission toggle (CUDA profiling only).

n_now

size_t ModuleBase::timer::n_now = 0

staticprivate

Member variable: the index of clocks.

timer_pool

std::map< std::string, std::map< std::string, timer::Timer_One > > ModuleBase::timer::timer_pool

static

---

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

• /home/runner/work/abacus-develop/abacus-develop/source/source_base/timer.h
• /home/runner/work/abacus-develop/abacus-develop/source/source_base/timer.cpp