This class is designed to abstract away the details of whether the project is built with MPI, or in single thread mode. More...

#include <ptems/parallelcomm.hpp>

Public Member Functions
	ParallelComm ()
	Initialise a reference to the "world" communicator. More...

	ParallelComm (ParallelComm &&rhs)
	Move constructor. More...

ParallelComm &	operator= (ParallelComm &&rhs)
	Move assignment. More...

	ParallelComm (const ParallelComm &rhs)
	Copy constructor. More...

ParallelComm &	operator= (const ParallelComm &rhs)
	Copy assignment. More...

	~ParallelComm ()
	Destroy the communicator object. More...

template<typename Dist >
auto	GenerateRandomNumber (Dist &dist)
	Generates a random number using the specified distribution. More...

void	DiscardRandomNumbers ([[maybe_unused]] unsigned long long amount)
	Discard the specified number of random numbers from the random number generator. More...

std::size_t	ProcessRank () const
	Gets the index of the current process in the communicator. More...

std::size_t	NumberProcessors () const
	Gets the number of processes in the communicator. More...

bool	IsMainProcess () const
	Gets whether the current process is the "main" process (first process in communicator) More...

template<typename T >
void	ThrowExceptionOnMain (const std::string &message) const
	Raises the specified exception on the main process (actually raises on all calling processes, but should only be called from code running only on main). More...

template<typename T >
void	CheckExceptionOnMain () const
	Checks if an exception was raised on the main process. More...

template<typename T = std::logic_error>
void	ThrowExceptionIfAny (const std::string &message, bool condition) const
	Raises the specified exception on all processors if the specified condition is true on ANY processor. More...

template<typename T = std::logic_error>
void	ThrowExceptionIfAll (const std::string &message, bool condition) const
	Raises the specified exception on all processors if the specified condition is true on ALL processors. More...

template<typename T >
std::enable_if< std::is_scalar< T >::value >::type	BroadcastFromMain (T &data) const
	Broadcast the specified value from the "main" process to all other processes. More...

template<typename T >
std::enable_if< std::is_scalar< T >::value >::type	BroadcastFrom (std::size_t processID, T &data) const
	Broadcast the specified value from the specified process to all other processes. More...

template<typename T >
auto	BroadcastFromMain ([[maybe_unused]] T &data) const -> decltype(data.data(), void())
	Broadcast the specified C++ random access container from the "main" process to all other processes (this covers std::string, std::vector, etc. More...

template<typename T >
auto	BroadcastFrom ([[maybe_unused]] std::size_t processID, [[maybe_unused]] T &data) const -> decltype(data.data(), void())
	Broadcast the specified C++ random access container from the specified process to all other processes (this covers std::string, std::vector, etc. More...

template<typename T >
void	BroadcastFromMain ([[maybe_unused]] T *data, [[maybe_unused]] std::size_t count) const
	Broadcast the specified array of data from the "main" process to all other processes. More...

template<typename T >
std::enable_if<!std::is_enum< T >::value >::type	BroadcastFrom ([[maybe_unused]] std::size_t processID, [[maybe_unused]] T *data, [[maybe_unused]] std::size_t count) const
	Broadcast the specified array of data from the specified process to all other processes. More...

template<typename T >
std::enable_if< std::is_enum< T >::value >::type	BroadcastFrom ([[maybe_unused]] std::size_t processID, [[maybe_unused]] T *data, [[maybe_unused]] std::size_t count) const
	Broadcast the specified array of data from the specified process to all other processes. More...

template<typename T >
void	Gather ([[maybe_unused]] T *data, [[maybe_unused]] std::size_t count) const
	Gather (concatenates) the specified number of elements from the array in each process into the array of the main (first) process. More...

template<typename T >
void	Gather ([[maybe_unused]] T *data, [[maybe_unused]] std::size_t offset, [[maybe_unused]] std::size_t count) const
	Gather the specified elements from the array in each process into the array of the main (first) process. More...

template<typename T >
void	Gather ([[maybe_unused]] T &data) const
	Gather (concatenates) the specified elements from the STL container in each process into the STL container of the main (first) process. More...

template<typename T >
void	AllGather (T data, std::size_t count, T out) const
	Gather (concatenates) the specified number of elements from the array in each process into the specified destination array of all processes. More...

template<typename T >
void	AllGather (T data, std::size_t offset, std::size_t count, T out) const
	Gather the specified elements from the array in each process into the destination array of all processes. More...

template<typename T >
void	AllGather ([[maybe_unused]] T &data) const
	Gather (concatenates) the specified elements from the STL container in each process into the STL container of all processes. More...

template<typename T >
void	Add (const T data, T output, std::size_t count) const
	Performs elementwise addition of the array from each process, and stores it into the destination array of the main process. More...

template<typename T >
void	AllAdd (const T data, T output, std::size_t count) const
	Performs elementwise addition of the array from each process, and stores it into the destination array of all processes. More...

template<typename T >
T	Add (T value) const
	Performs addition of the value from each process and returns the result on main process only. More...

template<typename T >
T	AllAdd (T value) const
	Performs addition of the value from each process and returns the value. More...

template<typename T >
T	Max (T value) const
	Performs maximum of the value from each process and returns the result on main process only. More...

template<typename T >
T	AllMax (T value) const
	Performs maximum of the value from each process and returns the value. More...

template<typename T >
T	Min (T value) const
	Performs minimum of the value from each process and returns the result on main process only. More...

template<typename T >
T	AllMin (T value) const
	Performs minimum of the value from each process and returns the value. More...

void	And (const bool data, bool output, std::size_t count) const
	Performs elementwise logical AND of the array from each process, and stores it into the destination array of the main process. More...

void	AllAnd (const bool data, bool output, std::size_t count) const
	Performs elementwise logical AND of the array from each process, and stores it into the destination array of all processes. More...

bool	And (bool value) const
	Performs elementwise logical AND of the boolean value from each process and returns the result on main process only. More...

bool	AllAnd (bool value) const
	Performs elementwise logical AND of the boolean value from each process and returns the value. More...

void	Or (const bool data, bool output, std::size_t count) const
	Performs elementwise logical AND of the array from each process, and stores it into the destination array of the main process. More...

void	AllOr (const bool data, bool output, std::size_t count) const
	Performs elementwise logical AND of the array from each process, and stores it into the destination array of all processes. More...

bool	Or (bool value) const
	Performs elementwise logical OR of the boolean value from each process and returns the result on main process only. More...

bool	AllOr (bool value) const
	Performs elementwise logical OR of the boolean value from each process and returns the value. More...

ParallelComm	Clone () const
	Creates a duplicate of the current communicator. More...

ParallelComm	Split (int colour, int key) const
	Splits the current communicator into multiple sub communicators. More...

Static Public Member Functions
static ParallelComm	World ()
	The main "world" communicator. More...

Detailed Description

This class is designed to abstract away the details of whether the project is built with MPI, or in single thread mode.

The general principle is to allow the rest of the code to use this class as the MPI communication assuming MPI is available. If, however, the application is built in single process mode this class "fakes" the MPI details (such as returning that only one process is available, the only process is the main process, and all gather/broadcast calls are no-ops).

It also manages a random number generator so that in a multiple processor system each processor uses the same seeded generator but SHOULD skip the generation of numbers generated by other processors

Constructor & Destructor Documentation

◆ ParallelComm() [1/3]

ptems::ParallelComm::ParallelComm ( )

Initialise a reference to the "world" communicator.

◆ ParallelComm() [2/3]

ptems::ParallelComm::ParallelComm ( ParallelComm && rhs )

Move constructor.

Parameters

rhs	The ParallelComm object to move to this

◆ ParallelComm() [3/3]

ptems::ParallelComm::ParallelComm ( const ParallelComm & rhs )

Copy constructor.

Parameters

rhs	The ParallelComm object to copy to this

◆ ~ParallelComm()

ptems::ParallelComm::~ParallelComm ( )

Destroy the communicator object.

Member Function Documentation

◆ Add() [1/2]

template<typename T >

void ptems::ParallelComm::Add	(	const T *	data,
		T *	output,
		std::size_t	count
	)		const

inline

Performs elementwise addition of the array from each process, and stores it into the destination array of the main process.

Template Parameters

T	The type of data to add

Parameters

data	Pointer to the array of data to add elementwise
output	Pointer to the destination array to store the result (main process only)
count	Number of elements in the input and output arrays

◆ Add() [2/2]

template<typename T >

T ptems::ParallelComm::Add ( T value ) const

inline

Performs addition of the value from each process and returns the result on main process only.

Parameters

value Value from current process to add

Returns: The sum of values from all processes (on main process); or just value on other processes (no addition performed)

◆ AllAdd() [1/2]

template<typename T >

void ptems::ParallelComm::AllAdd	(	const T *	data,
		T *	output,
		std::size_t	count
	)		const

inline

Performs elementwise addition of the array from each process, and stores it into the destination array of all processes.

Template Parameters

T	The type of data to add

Parameters

data	Pointer to the array of data to add elementwise
output	Pointer to the destination array to store the result
count	Number of elements in the input and output arrays

◆ AllAdd() [2/2]

template<typename T >

T ptems::ParallelComm::AllAdd ( T value ) const

inline

Performs addition of the value from each process and returns the value.

Parameters

value Value from current process to add

Returns: The sum of values from all processes.

◆ AllAnd() [1/2]

bool ptems::ParallelComm::AllAnd ( bool value ) const

inline

Performs elementwise logical AND of the boolean value from each process and returns the value.

Parameters

value Boolean value from current process to AND

Returns: The logical AND of values from all processes.

◆ AllAnd() [2/2]

void ptems::ParallelComm::AllAnd	(	const bool *	data,
		bool *	output,
		std::size_t	count
	)		const

inline

Performs elementwise logical AND of the array from each process, and stores it into the destination array of all processes.

Parameters

data	Pointer to the array of data to AND elementwise
output	Pointer to the destination array to store the result
count	Number of elements in the input and output arrays

◆ AllGather() [1/3]

template<typename T >

void ptems::ParallelComm::AllGather ( [[maybe_unused] ] T & data ) const

inline

Gather (concatenates) the specified elements from the STL container in each process into the STL container of all processes.

The number of entries in each process can be different.

Wrapper around MPI_Allgatherv

Template Parameters

T	The type of the STL container to gather

Parameters

data	The STL container containing data to gather. On the main process the gathered data is placed into this container.

◆ AllGather() [2/3]

template<typename T >

void ptems::ParallelComm::AllGather	(	T *	data,
		std::size_t	count,
		T *	out
	)		const

inline

Gather (concatenates) the specified number of elements from the array in each process into the specified destination array of all processes.

Wrapper around MPI_Allgather

Template Parameters

T	The type of data to gather

Parameters

data	Pointer to the array of data to gather.
count	The number of elements to send. Must be the same for all processes.
out	Array to gather the data into. The array in the main process MUST be large enough to store the data from all processes. This array can be the same as data.

◆ AllGather() [3/3]

template<typename T >

void ptems::ParallelComm::AllGather	(	T *	data,
		std::size_t	offset,
		std::size_t	count,
		T *	out
	)		const

inline

Gather the specified elements from the array in each process into the destination array of all processes.

The number of entries in each process can be different, and can be gathered into an arbitrary offset in the destination.

Wrapper around MPI_Allgatherv

Template Parameters

T	The type of data to gather

Parameters

data	Pointer to the array of data to gather.
offset	Offset into the destination array of the location to gather the data from this process
count	The number of elements to send. Can be different for different processes.
out	Array to gather the data into. The array in the main process MUST be large enough to store the data from all processes. This array can be the same as data.

◆ AllMax()

template<typename T >

T ptems::ParallelComm::AllMax ( T value ) const

inline

Performs maximum of the value from each process and returns the value.

Parameters

value Value from current process to add

Returns: The maximum of values from all processes.

◆ AllMin()

template<typename T >

T ptems::ParallelComm::AllMin ( T value ) const

inline

Performs minimum of the value from each process and returns the value.

Parameters

value Value from current process to add

Returns: The minimum of values from all processes.

◆ AllOr() [1/2]

bool ptems::ParallelComm::AllOr ( bool value ) const

inline

Performs elementwise logical OR of the boolean value from each process and returns the value.

Parameters

value Boolean value from current process to OR

Returns: The logical OR of values from all processes.

◆ AllOr() [2/2]

void ptems::ParallelComm::AllOr	(	const bool *	data,
		bool *	output,
		std::size_t	count
	)		const

inline

Performs elementwise logical AND of the array from each process, and stores it into the destination array of all processes.

Parameters

data	Pointer to the array of data to AND elementwise
output	Pointer to the destination array to store the result
count	Number of elements in the input and output arrays

◆ And() [1/2]

bool ptems::ParallelComm::And ( bool value ) const

inline

Performs elementwise logical AND of the boolean value from each process and returns the result on main process only.

Parameters

value Boolean value from current process to AND

Returns: The logical AND of values from all processes (on main process); or just value on other processes (no logical AND performed)

◆ And() [2/2]

void ptems::ParallelComm::And	(	const bool *	data,
		bool *	output,
		std::size_t	count
	)		const

inline

Performs elementwise logical AND of the array from each process, and stores it into the destination array of the main process.

Parameters

data	Pointer to the array of data to AND elementwise
output	Pointer to the destination array to store the result (main process only)
count	Number of elements in the input and output arrays

◆ BroadcastFrom() [1/4]

template<typename T >

auto ptems::ParallelComm::BroadcastFrom	(	[[maybe_unused] ] std::size_t	processID,
		[[maybe_unused] ] T &	data
	)		const -> decltype(data.data(), void())

inline

Broadcast the specified C++ random access container from the specified process to all other processes (this covers std::string, std::vector, etc.

but not std::list etc.)

Template Parameters

T	The container type to broadcast

Parameters

processID	The process ID to broadcast from
data	Data to broadcast

◆ BroadcastFrom() [2/4]

template<typename T >

std::enable_if<!std::is_enum<T>::value>::type ptems::ParallelComm::BroadcastFrom	(	[[maybe_unused] ] std::size_t	processID,
		[[maybe_unused] ] T *	data,
		[[maybe_unused] ] std::size_t	count
	)		const

inline

Broadcast the specified array of data from the specified process to all other processes.

Template Parameters

T	The type of data to broadcast, where T is NOT an enumeration type

Parameters

processID	The process ID to broadcast from
data	Pointer to the array of data to broadcast
count	The size of the array

◆ BroadcastFrom() [3/4]

template<typename T >

std::enable_if<std::is_enum<T>::value>::type ptems::ParallelComm::BroadcastFrom	(	[[maybe_unused] ] std::size_t	processID,
		[[maybe_unused] ] T *	data,
		[[maybe_unused] ] std::size_t	count
	)		const

inline

Broadcast the specified array of data from the specified process to all other processes.

Template Parameters

T	The type of data to broadcast, where T is an enumeration type

Parameters

processID	The process ID to broadcast from
data	Pointer to the array of data to broadcast
count	The size of the array

◆ BroadcastFrom() [4/4]

template<typename T >

std::enable_if<std::is_scalar<T>::value>::type ptems::ParallelComm::BroadcastFrom	(	std::size_t	processID,
		T &	data
	)		const

inline

Broadcast the specified value from the specified process to all other processes.

Template Parameters

T	The type of data to broadcast

Parameters

processID	The process ID to broadcast from
data	Data to broadcast

◆ BroadcastFromMain() [1/3]

template<typename T >

auto ptems::ParallelComm::BroadcastFromMain ( [[maybe_unused] ] T & data ) const -> decltype(data.data(), void())

inline

Broadcast the specified C++ random access container from the "main" process to all other processes (this covers std::string, std::vector, etc.

but not std::list etc.)

Template Parameters

T	The container type to broadcast

Parameters

data	Data to broadcast

◆ BroadcastFromMain() [2/3]

template<typename T >

void ptems::ParallelComm::BroadcastFromMain	(	[[maybe_unused] ] T *	data,
		[[maybe_unused] ] std::size_t	count
	)		const

inline

Broadcast the specified array of data from the "main" process to all other processes.

Template Parameters

T	The type of data to broadcast, where T is NOT an enumeration type

Parameters

data	Pointer to the array of data to broadcast
count	The size of the array

◆ BroadcastFromMain() [3/3]

template<typename T >

std::enable_if<std::is_scalar<T>::value>::type ptems::ParallelComm::BroadcastFromMain ( T & data ) const

inline

Broadcast the specified value from the "main" process to all other processes.

Template Parameters

T	The type of data to broadcast

Parameters

data	Data to broadcast

◆ CheckExceptionOnMain()

template<typename T >

void ptems::ParallelComm::CheckExceptionOnMain ( ) const

inline

Checks if an exception was raised on the main process.

Should be called on ALL processes (even main if no exception was raised) after any block of code that may have called ThrowExceptionOnMain

Template Parameters

T	The exception type to raise is an exception occurred on main

Exceptions

T	(which may not be the same as the type raised by ThrowExceptionOnMain) if an exception occurred on main

◆ Clone()

ParallelComm ptems::ParallelComm::Clone ( ) const

Creates a duplicate of the current communicator.

Returns: The duplicated communicator.

◆ DiscardRandomNumbers()

void ptems::ParallelComm::DiscardRandomNumbers ( [[maybe_unused] ] unsigned long long amount )

inline

Discard the specified number of random numbers from the random number generator.

◆ Gather() [1/3]

template<typename T >

void ptems::ParallelComm::Gather ( [[maybe_unused] ] T & data ) const

inline

Gather (concatenates) the specified elements from the STL container in each process into the STL container of the main (first) process.

The number of entries in each process can be different.

Wrapper around MPI_Gatherv

Template Parameters

T	The type of the STL container to gather

Parameters

data	The STL container containing data to gather. On the main process the gathered data is placed into this container.

◆ Gather() [2/3]

template<typename T >

void ptems::ParallelComm::Gather	(	[[maybe_unused] ] T *	data,
		[[maybe_unused] ] std::size_t	count
	)		const

inline

Gather (concatenates) the specified number of elements from the array in each process into the array of the main (first) process.

Wrapper around MPI_Gather

Template Parameters

T	The type of data to gather

Parameters

data	Pointer to the array of data to gather. On the main process the gathered data is placed into this array (the main process should already have the data in the first count entries). The array in the main process MUST be large enough to store the data from all processes.
count	The number of elements to send. Must be the same for all processes.

◆ Gather() [3/3]

template<typename T >

void ptems::ParallelComm::Gather	(	[[maybe_unused] ] T *	data,
		[[maybe_unused] ] std::size_t	offset,
		[[maybe_unused] ] std::size_t	count
	)		const

inline

Gather the specified elements from the array in each process into the array of the main (first) process.

The number of entries in each process can be different, and can be gathered into an arbitrary offset in the main process.

Wrapper around MPI_Gatherv

Template Parameters

T	The type of data to gather

Parameters

data	Pointer to the array of data to gather. On the main process the gathered data is placed into this array (the main process should already have the data in the correct location). The array in the main process MUST be large enough to store the data from all processes.
offset	Offset into the array in the main process of the location to gather the data from this process
count	The number of elements to send. Can be different for different processes.

◆ GenerateRandomNumber()

template<typename Dist >

auto ptems::ParallelComm::GenerateRandomNumber ( Dist & dist )

inline

Generates a random number using the specified distribution.

Note: This method assumes that each processor calls this function the same number of times. This may not be 100% accurate, which may cause issues with the PNG properties. As such, possibly best to at least call DiscardRandomNumbers() to attempt to even the calls

Template Parameters

The	type of the random number distribution

Parameters

dist	The random number distribution to use

Returns: The generated random number

◆ IsMainProcess()

bool ptems::ParallelComm::IsMainProcess ( ) const

inline

Gets whether the current process is the "main" process (first process in communicator)

Returns: true If the current process is the first processor in the communicator; false If the current process is not the first processor in communicator

◆ Max()

template<typename T >

T ptems::ParallelComm::Max ( T value ) const

inline

Performs maximum of the value from each process and returns the result on main process only.

Parameters

value Value from current process to calculate max of

Returns: The maximum of values from all processes (on main process); or just value on other processes (no max performed)

◆ Min()

template<typename T >

T ptems::ParallelComm::Min ( T value ) const

inline

Performs minimum of the value from each process and returns the result on main process only.

Parameters

value Value from current process to calculate min of

Returns: The minimum of values from all processes (on main process); or just value on other processes (no min performed)

◆ NumberProcessors()

std::size_t ptems::ParallelComm::NumberProcessors ( ) const

inline

Gets the number of processes in the communicator.

Returns: Number of processes

◆ operator=() [1/2]

ParallelComm& ptems::ParallelComm::operator= ( const ParallelComm & rhs )

Copy assignment.

Parameters

rhs	The ParallelComm object to move to this

◆ operator=() [2/2]

ParallelComm& ptems::ParallelComm::operator= ( ParallelComm && rhs )

Move assignment.

Parameters

rhs	The ParallelComm object to move to this

◆ Or() [1/2]

bool ptems::ParallelComm::Or ( bool value ) const

inline

Performs elementwise logical OR of the boolean value from each process and returns the result on main process only.

Parameters

value Boolean value from current process to OR

Returns: The logical OR of values from all processes (on main process); or just value on other processes (no logical OR performed)

◆ Or() [2/2]

void ptems::ParallelComm::Or	(	const bool *	data,
		bool *	output,
		std::size_t	count
	)		const

inline

Performs elementwise logical AND of the array from each process, and stores it into the destination array of the main process.

Parameters

data	Pointer to the array of data to AND elementwise
output	Pointer to the destination array to store the result (main process only)
count	Number of elements in the input and output arrays

◆ ProcessRank()

std::size_t ptems::ParallelComm::ProcessRank ( ) const

inline

Gets the index of the current process in the communicator.

Returns: Index of current process

◆ Split()

ParallelComm ptems::ParallelComm::Split	(	int	colour,
		int	key
	)		const

Splits the current communicator into multiple sub communicators.

Wrapper around MPI_Comm_split. All processes which passes the same value for colour will end up in the subcommunicator, ordered such that the lowest key of all processes in the same communicator is the "first" process.

Parameters

colour	Index denoting subcommunicator process will be added to
key	Key to use to indicate order of processes in new communicators

Returns: The subcommunicator containing this process

◆ ThrowExceptionIfAll()

template<typename T = std::logic_error>

void ptems::ParallelComm::ThrowExceptionIfAll	(	const std::string &	message,
		bool	condition
	)		const

inline

Raises the specified exception on all processors if the specified condition is true on ALL processors.

Template Parameters

T	The exception type to raise

Parameters

message	The exception error message
condition	Condition that if true on ALL processors causes the excpetion to be thrown

Exceptions

T	if condition is true on ALL process

◆ ThrowExceptionIfAny()

template<typename T = std::logic_error>

void ptems::ParallelComm::ThrowExceptionIfAny	(	const std::string &	message,
		bool	condition
	)		const

inline

Raises the specified exception on all processors if the specified condition is true on ANY processor.

Template Parameters

T	The exception type to raise

Parameters

message	The exception error message
condition	Condition that if true on ANY processors causes the excpetion to be thrown

Exceptions

T	if condition is true on ALL process

◆ ThrowExceptionOnMain()

template<typename T >

void ptems::ParallelComm::ThrowExceptionOnMain ( const std::string & message ) const

inline

Raises the specified exception on the main process (actually raises on all calling processes, but should only be called from code running only on main).

After any block of code which calls ThrowExceptionOnMain a call to CheckExceptionOnMain should be made on ALL processes (including main process if no exception was raised)

Template Parameters

T	The exception type to raise

Parameters

message The exception error message

Exceptions

T	is always thrown

◆ World()

static ParallelComm ptems::ParallelComm::World ( )

static

The main "world" communicator.

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

ptems/parallelcomm.hpp

Public Member Functions

Static Public Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ ParallelComm() [1/3]

◆ ParallelComm() [2/3]

◆ ParallelComm() [3/3]

◆ ~ParallelComm()

Member Function Documentation

◆ Add() [1/2]

◆ Add() [2/2]

◆ AllAdd() [1/2]

◆ AllAdd() [2/2]

◆ AllAnd() [1/2]

◆ AllAnd() [2/2]

◆ AllGather() [1/3]

◆ AllGather() [2/3]

◆ AllGather() [3/3]

◆ AllMax()

◆ AllMin()

◆ AllOr() [1/2]

◆ AllOr() [2/2]

◆ And() [1/2]

◆ And() [2/2]

◆ BroadcastFrom() [1/4]

◆ BroadcastFrom() [2/4]

◆ BroadcastFrom() [3/4]

◆ BroadcastFrom() [4/4]

◆ BroadcastFromMain() [1/3]

◆ BroadcastFromMain() [2/3]

◆ BroadcastFromMain() [3/3]

◆ CheckExceptionOnMain()

◆ Clone()

◆ DiscardRandomNumbers()

◆ Gather() [1/3]

◆ Gather() [2/3]

◆ Gather() [3/3]

◆ GenerateRandomNumber()

◆ IsMainProcess()

◆ Max()

◆ Min()

◆ NumberProcessors()

◆ operator=() [1/2]

◆ operator=() [2/2]

◆ Or() [1/2]

◆ Or() [2/2]

◆ ProcessRank()

◆ Split()

◆ ThrowExceptionIfAll()

◆ ThrowExceptionIfAny()

◆ ThrowExceptionOnMain()

◆ World()