monolish::mpi::comm Class Reference

MPI class (singleton) More...

#include <monolish_mpi_core.hpp>

Collaboration diagram for monolish::mpi::comm:

Public Member Functions
double	Allreduce (double val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
float	Allreduce (float val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
int	Allreduce (int val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
size_t	Allreduce (size_t val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
double	Allreduce_max (double val) const
	MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes. More...
float	Allreduce_max (float val) const
	MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes. More...
int	Allreduce_max (int val) const
	MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes. More...
size_t	Allreduce_max (size_t val) const
	MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes. More...
double	Allreduce_min (double val) const
	MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes. More...
float	Allreduce_min (float val) const
	MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes. More...
int	Allreduce_min (int val) const
	MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes. More...
size_t	Allreduce_min (size_t val) const
	MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes. More...
double	Allreduce_prod (double val) const
	MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes. More...
float	Allreduce_prod (float val) const
	MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes. More...
int	Allreduce_prod (int val) const
	MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes. More...
size_t	Allreduce_prod (size_t val) const
	MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes. More...
double	Allreduce_sum (double val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
float	Allreduce_sum (float val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
int	Allreduce_sum (int val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
size_t	Allreduce_sum (size_t val) const
	MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes. More...
void	Barrier () const
	Blocks until all processes in the communicator have reached this routine. More...
void	Bcast (double &val, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (float &val, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (int &val, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (monolish::vector< double > &vec, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (monolish::vector< float > &vec, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (size_t &val, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (std::vector< double > &vec, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (std::vector< float > &vec, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (std::vector< int > &vec, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
void	Bcast (std::vector< size_t > &vec, int root) const
	MPI_Bcast, Broadcasts a message from the process with rank root to all other processes. More...
	comm (comm &&)=delete
	comm (const comm &)=delete
void	Finalize ()
	Terminates MPI execution environment. More...
void	Gather (monolish::vector< double > &sendvec, monolish::vector< double > &recvvec, int root) const
	MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process. More...
void	Gather (monolish::vector< float > &sendvec, monolish::vector< float > &recvvec, int root) const
	MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process. More...
void	Gather (std::vector< double > &sendvec, std::vector< double > &recvvec, int root) const
	MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process. More...
void	Gather (std::vector< float > &sendvec, std::vector< float > &recvvec, int root) const
	MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process. More...
void	Gather (std::vector< int > &sendvec, std::vector< int > &recvvec, int root) const
	MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process. More...
void	Gather (std::vector< size_t > &sendvec, std::vector< size_t > &recvvec, int root) const
	MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process. More...
MPI_Comm	get_comm () const
	get communicator More...
int	get_rank ()
	get my rank number More...
int	get_size ()
	get the number of processes More...
void	Init ()
	Initialize the MPI execution environment. More...
void	Init (int argc, char **argv)
	Initialize the MPI execution environment. More...
bool	Initialized () const
	Indicates whether MPI_Init has been called. More...
void	Irecv (double val, int src, int tag)
	MPI_Irecv for scalar. Performs a nonblocking recv. More...
void	Irecv (float val, int src, int tag)
	MPI_Irecv for scalar. Performs a nonblocking recv. More...
void	Irecv (int val, int src, int tag)
	MPI_Irecv for scalar. Performs a nonblocking recv. More...
void	Irecv (monolish::vector< double > &vec, int src, int tag)
	MPI_Irecv for monolish::vector. Performs a nonblocking recv. More...
void	Irecv (monolish::vector< float > &vec, int src, int tag)
	MPI_Irecv for monolish::vector. Performs a nonblocking recv. More...
void	Irecv (size_t val, int src, int tag)
	MPI_Irecv for scalar. Performs a nonblocking recv. More...
void	Irecv (std::vector< double > &vec, int src, int tag)
	MPI_Irecv for std::vector. Performs a nonblocking recv. More...
void	Irecv (std::vector< float > &vec, int src, int tag)
	MPI_Irecv for std::vector. Performs a nonblocking recv. More...
void	Irecv (std::vector< int > &vec, int src, int tag)
	MPI_Irecv for std::vector. Performs a nonblocking recv. More...
void	Irecv (std::vector< size_t > &vec, int src, int tag)
	MPI_Irecv for std::vector. Performs a nonblocking recv. More...
void	Isend (const monolish::vector< double > &vec, int dst, int tag)
	MPI_Isend for monolish::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (const monolish::vector< float > &vec, int dst, int tag)
	MPI_Isend for monolish::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (const std::vector< double > &vec, int dst, int tag)
	MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (const std::vector< float > &vec, int dst, int tag)
	MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (const std::vector< int > &vec, int dst, int tag)
	MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (const std::vector< size_t > &vec, int dst, int tag)
	MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (double val, int dst, int tag)
	MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (float val, int dst, int tag)
	MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (int val, int dst, int tag)
	MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
void	Isend (size_t val, int dst, int tag)
	MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall(). More...
comm &	operator= (comm &&)=delete
comm &	operator= (const comm &)=delete
MPI_Status	Recv (double val, int src, int tag) const
	MPI_Recv for scalar. Performs a blocking recv. More...
MPI_Status	Recv (float val, int src, int tag) const
	MPI_Recv for scalar. Performs a blocking recv. More...
MPI_Status	Recv (int val, int src, int tag) const
	MPI_Recv for scalar. Performs a blocking recv. More...
MPI_Status	Recv (monolish::vector< double > &vec, int src, int tag) const
	MPI_Recv for monolish::vector. Performs a blocking recv. More...
MPI_Status	Recv (monolish::vector< float > &vec, int src, int tag) const
	MPI_Recv for monolish::vector. Performs a blocking recv. More...
MPI_Status	Recv (size_t val, int src, int tag) const
	MPI_Recv for scalar. Performs a blocking recv. More...
MPI_Status	Recv (std::vector< double > &vec, int src, int tag) const
	MPI_Recv for std::vector. Performs a blocking recv. More...
MPI_Status	Recv (std::vector< float > &vec, int src, int tag) const
	MPI_Recv for std::vector. Performs a blocking recv. More...
MPI_Status	Recv (std::vector< int > &vec, int src, int tag) const
	MPI_Recv for std::vector. Performs a blocking recv. More...
MPI_Status	Recv (std::vector< size_t > &vec, int src, int tag) const
	MPI_Recv for std::vector. Performs a blocking recv. More...
void	Scatter (monolish::vector< double > &sendvec, monolish::vector< double > &recvvec, int root) const
	MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process. More...
void	Scatter (monolish::vector< float > &sendvec, monolish::vector< float > &recvvec, int root) const
	MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process. More...
void	Scatter (std::vector< double > &sendvec, std::vector< double > &recvvec, int root) const
	MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process. More...
void	Scatter (std::vector< float > &sendvec, std::vector< float > &recvvec, int root) const
	MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process. More...
void	Scatter (std::vector< int > &sendvec, std::vector< int > &recvvec, int root) const
	MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process. More...
void	Scatter (std::vector< size_t > &sendvec, std::vector< size_t > &recvvec, int root) const
	MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process. More...
void	Send (double val, int dst, int tag) const
	MPI_Send for scalar. Performs a blocking send. More...
void	Send (float val, int dst, int tag) const
	MPI_Send for scalar. Performs a blocking send. More...
void	Send (int val, int dst, int tag) const
	MPI_Send for scalar. Performs a blocking send. More...
void	Send (monolish::vector< double > &vec, int dst, int tag) const
	MPI_Send for monolish::vector. Performs a blocking send. More...
void	Send (monolish::vector< float > &vec, int dst, int tag) const
	MPI_Send for monolish::vector. Performs a blocking send. More...
void	Send (size_t val, int dst, int tag) const
	MPI_Send for scalar. Performs a blocking send. More...
void	Send (std::vector< double > &vec, int dst, int tag) const
	MPI_Send for std::vector. Performs a blocking send. More...
void	Send (std::vector< float > &vec, int dst, int tag) const
	MPI_Send for std::vector. Performs a blocking send. More...
void	Send (std::vector< int > &vec, int dst, int tag) const
	MPI_Send for std::vector. Performs a blocking send. More...
void	Send (std::vector< size_t > &vec, int dst, int tag) const
	MPI_Send for std::vector. Performs a blocking send. More...
void	set_comm (MPI_Comm external_comm)
	set communicator More...
void	Waitall ()
	Waits for all communications to complete. More...

Static Public Member Functions
static comm &	get_instance ()

Private Member Functions
	comm ()
	~comm ()

Private Attributes
MPI_Comm	my_comm = 0
	MPI communicator, MPI_COMM_WORLD. More...
std::vector< MPI_Request >	requests

Detailed Description

MPI class (singleton)

Definition at line 22 of file monolish_mpi_core.hpp.

Constructor & Destructor Documentation

comm() [1/3]

monolish::mpi::comm::comm ( )

inlineprivate

Definition at line 28 of file monolish_mpi_core.hpp.

~comm()

monolish::mpi::comm::~comm ( )

inlineprivate

Definition at line 29 of file monolish_mpi_core.hpp.

comm() [2/3]

monolish::mpi::comm::comm ( const comm &  )

delete

comm() [3/3]

monolish::mpi::comm::comm ( comm &&  )

delete

Member Function Documentation

Allreduce() [1/4]

double monolish::mpi::comm::Allreduce

(

double

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce() [2/4]

float monolish::mpi::comm::Allreduce

(

float

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce() [3/4]

int monolish::mpi::comm::Allreduce

(

int

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce() [4/4]

size_t monolish::mpi::comm::Allreduce

(

size_t

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_max() [1/4]

double monolish::mpi::comm::Allreduce_max

(

double

val

)

const

MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_max() [2/4]

float monolish::mpi::comm::Allreduce_max

(

float

val

)

const

MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_max() [3/4]

int monolish::mpi::comm::Allreduce_max

(

int

val

)

const

MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_max() [4/4]

size_t monolish::mpi::comm::Allreduce_max

(

size_t

val

)

const

MPI_Allreduce (MPI_MAX) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_min() [1/4]

double monolish::mpi::comm::Allreduce_min

(

double

val

)

const

MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_min() [2/4]

float monolish::mpi::comm::Allreduce_min

(

float

val

)

const

MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_min() [3/4]

int monolish::mpi::comm::Allreduce_min

(

int

val

)

const

MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_min() [4/4]

size_t monolish::mpi::comm::Allreduce_min

(

size_t

val

)

const

MPI_Allreduce (MPI_MIN) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_prod() [1/4]

double monolish::mpi::comm::Allreduce_prod

(

double

val

)

const

MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_prod() [2/4]

float monolish::mpi::comm::Allreduce_prod

(

float

val

)

const

MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_prod() [3/4]

int monolish::mpi::comm::Allreduce_prod

(

int

val

)

const

MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_prod() [4/4]

size_t monolish::mpi::comm::Allreduce_prod

(

size_t

val

)

const

MPI_Allreduce (MPI_PROD) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_sum() [1/4]

double monolish::mpi::comm::Allreduce_sum

(

double

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_sum() [2/4]

float monolish::mpi::comm::Allreduce_sum

(

float

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_sum() [3/4]

int monolish::mpi::comm::Allreduce_sum

(

int

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Allreduce_sum() [4/4]

size_t monolish::mpi::comm::Allreduce_sum

(

size_t

val

)

const

MPI_Allreduce (MPI_SUM) for scalar. Combines values from all processes and distributes the result back to all processes.

Parameters

val	scalar value

Barrier()

void monolish::mpi::comm::Barrier

(

)

const

Blocks until all processes in the communicator have reached this routine.

Bcast() [1/10]

void monolish::mpi::comm::Bcast	(	double &	val,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

val	scalar value
root	root rank number

Bcast() [2/10]

void monolish::mpi::comm::Bcast	(	float &	val,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

val	scalar value
root	root rank number

Bcast() [3/10]

void monolish::mpi::comm::Bcast	(	int &	val,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

val	scalar value
root	root rank number

Bcast() [4/10]

void monolish::mpi::comm::Bcast	(	monolish::vector< double > &	vec,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

vec	monolish vector (size N)
root	root rank number

Bcast() [5/10]

void monolish::mpi::comm::Bcast	(	monolish::vector< float > &	vec,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

vec	monolish vector (size N)
root	root rank number

Bcast() [6/10]

void monolish::mpi::comm::Bcast	(	size_t &	val,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

val	scalar value
root	root rank number

Bcast() [7/10]

void monolish::mpi::comm::Bcast	(	std::vector< double > &	vec,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

vec	std::vector (size N)
root	root rank number

Bcast() [8/10]

void monolish::mpi::comm::Bcast	(	std::vector< float > &	vec,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

vec	std::vector (size N)
root	root rank number

Bcast() [9/10]

void monolish::mpi::comm::Bcast	(	std::vector< int > &	vec,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

vec	std::vector (size N)
root	root rank number

Bcast() [10/10]

void monolish::mpi::comm::Bcast	(	std::vector< size_t > &	vec,
		int	root
	)		const

MPI_Bcast, Broadcasts a message from the process with rank root to all other processes.

Parameters

vec	std::vector (size N)
root	root rank number

Finalize()

void monolish::mpi::comm::Finalize

(

)

Terminates MPI execution environment.

Gather() [1/6]

void monolish::mpi::comm::Gather	(	monolish::vector< double > &	sendvec,
		monolish::vector< double > &	recvvec,
		int	root
	)		const

MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, monolish vector (size N)
recvvec	recv data, monolish vector (size N * # of procs)
root	root rank number

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Gather() [2/6]

void monolish::mpi::comm::Gather	(	monolish::vector< float > &	sendvec,
		monolish::vector< float > &	recvvec,
		int	root
	)		const

MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, monolish vector (size N)
recvvec	recv data, monolish vector (size N * # of procs)
root	root rank number

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Gather() [3/6]

void monolish::mpi::comm::Gather	(	std::vector< double > &	sendvec,
		std::vector< double > &	recvvec,
		int	root
	)		const

MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std vector (size N)
recvvec	recv data, std vector (size N * # of procs)
root	root rank number

Gather() [4/6]

void monolish::mpi::comm::Gather	(	std::vector< float > &	sendvec,
		std::vector< float > &	recvvec,
		int	root
	)		const

MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std vector (size N)
recvvec	recv data, std vector (size N * # of procs)
root	root rank number

Gather() [5/6]

void monolish::mpi::comm::Gather	(	std::vector< int > &	sendvec,
		std::vector< int > &	recvvec,
		int	root
	)		const

MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std vector (size N)
recvvec	recv data, std vector (size N * # of procs)
root	root rank number

Gather() [6/6]

void monolish::mpi::comm::Gather	(	std::vector< size_t > &	sendvec,
		std::vector< size_t > &	recvvec,
		int	root
	)		const

MPI_Gather, Gathers vector from all processes The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std vector (size N)
recvvec	recv data, std vector (size N * # of procs)
root	root rank number

get_comm()

MPI_Comm monolish::mpi::comm::get_comm ( ) const

inline

get communicator

Returns

MPI_COMM_WORLD

Definition at line 68 of file monolish_mpi_core.hpp.

get_instance()

static comm& monolish::mpi::comm::get_instance ( )

inlinestatic

Definition at line 39 of file monolish_mpi_core.hpp.

get_rank()

int monolish::mpi::comm::get_rank

(

)

get my rank number

Returns

rank number

get_size()

int monolish::mpi::comm::get_size

(

)

get the number of processes

Returns

the number of prodessed

Init() [1/2]

void monolish::mpi::comm::Init

(

)

Initialize the MPI execution environment.

Init() [2/2]

void monolish::mpi::comm::Init	(	int	argc,
		char **	argv
	)

Initialize the MPI execution environment.

Parameters

argc	Pointer to the number of arguments
argv	Pointer to the argument vector

Initialized()

bool monolish::mpi::comm::Initialized

(

)

const

Indicates whether MPI_Init has been called.

Returns

true: initialized, false: not initialized

Irecv() [1/10]

void monolish::mpi::comm::Irecv	(	double	val,
		int	src,
		int	tag
	)

MPI_Irecv for scalar. Performs a nonblocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [2/10]

void monolish::mpi::comm::Irecv	(	float	val,
		int	src,
		int	tag
	)

MPI_Irecv for scalar. Performs a nonblocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [3/10]

void monolish::mpi::comm::Irecv	(	int	val,
		int	src,
		int	tag
	)

MPI_Irecv for scalar. Performs a nonblocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [4/10]

void monolish::mpi::comm::Irecv	(	monolish::vector< double > &	vec,
		int	src,
		int	tag
	)

MPI_Irecv for monolish::vector. Performs a nonblocking recv.

Parameters

vec	monolish::vector (size N)
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function. This function is not thread-safe.

Irecv() [5/10]

void monolish::mpi::comm::Irecv	(	monolish::vector< float > &	vec,
		int	src,
		int	tag
	)

MPI_Irecv for monolish::vector. Performs a nonblocking recv.

Parameters

vec	monolish::vector (size N)
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function. This function is not thread-safe.

Irecv() [6/10]

void monolish::mpi::comm::Irecv	(	size_t	val,
		int	src,
		int	tag
	)

MPI_Irecv for scalar. Performs a nonblocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [7/10]

void monolish::mpi::comm::Irecv	(	std::vector< double > &	vec,
		int	src,
		int	tag
	)

MPI_Irecv for std::vector. Performs a nonblocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [8/10]

void monolish::mpi::comm::Irecv	(	std::vector< float > &	vec,
		int	src,
		int	tag
	)

MPI_Irecv for std::vector. Performs a nonblocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [9/10]

void monolish::mpi::comm::Irecv	(	std::vector< int > &	vec,
		int	src,
		int	tag
	)

MPI_Irecv for std::vector. Performs a nonblocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Irecv() [10/10]

void monolish::mpi::comm::Irecv	(	std::vector< size_t > &	vec,
		int	src,
		int	tag
	)

MPI_Irecv for std::vector. Performs a nonblocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [1/10]

void monolish::mpi::comm::Isend	(	const monolish::vector< double > &	vec,
		int	dst,
		int	tag
	)

MPI_Isend for monolish::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function. This function is not thread-safe.

Isend() [2/10]

void monolish::mpi::comm::Isend	(	const monolish::vector< float > &	vec,
		int	dst,
		int	tag
	)

MPI_Isend for monolish::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function. This function is not thread-safe.

Isend() [3/10]

void monolish::mpi::comm::Isend	(	const std::vector< double > &	vec,
		int	dst,
		int	tag
	)

MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [4/10]

void monolish::mpi::comm::Isend	(	const std::vector< float > &	vec,
		int	dst,
		int	tag
	)

MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [5/10]

void monolish::mpi::comm::Isend	(	const std::vector< int > &	vec,
		int	dst,
		int	tag
	)

MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [6/10]

void monolish::mpi::comm::Isend	(	const std::vector< size_t > &	vec,
		int	dst,
		int	tag
	)

MPI_Isend for std::vector. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [7/10]

void monolish::mpi::comm::Isend	(	double	val,
		int	dst,
		int	tag
	)

MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [8/10]

void monolish::mpi::comm::Isend	(	float	val,
		int	dst,
		int	tag
	)

MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [9/10]

void monolish::mpi::comm::Isend	(	int	val,
		int	dst,
		int	tag
	)

MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

Isend() [10/10]

void monolish::mpi::comm::Isend	(	size_t	val,
		int	dst,
		int	tag
	)

MPI_Isend for scalar. Performs a nonblocking send. Requests are stored internally. All requests are synchronized by Waitall().

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Note

There is not MPI_Wait() in monolish::mpi, all communication is synchronized by using Waitall() function.

Warning

This function is not thread-safe.

operator=() [1/2]

comm& monolish::mpi::comm::operator= ( comm &&  )

delete

operator=() [2/2]

comm& monolish::mpi::comm::operator= ( const comm &  )

delete

Recv() [1/10]

MPI_Status monolish::mpi::comm::Recv	(	double	val,
		int	src,
		int	tag
	)		const

MPI_Recv for scalar. Performs a blocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [2/10]

MPI_Status monolish::mpi::comm::Recv	(	float	val,
		int	src,
		int	tag
	)		const

MPI_Recv for scalar. Performs a blocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [3/10]

MPI_Status monolish::mpi::comm::Recv	(	int	val,
		int	src,
		int	tag
	)		const

MPI_Recv for scalar. Performs a blocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [4/10]

MPI_Status monolish::mpi::comm::Recv	(	monolish::vector< double > &	vec,
		int	src,
		int	tag
	)		const

MPI_Recv for monolish::vector. Performs a blocking recv.

Parameters

vec	monolish::vector (size N)
src	rank of source
tag	message tag

Returns

MPI status object

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Recv() [5/10]

MPI_Status monolish::mpi::comm::Recv	(	monolish::vector< float > &	vec,
		int	src,
		int	tag
	)		const

MPI_Recv for monolish::vector. Performs a blocking recv.

Parameters

vec	monolish::vector (size N)
src	rank of source
tag	message tag

Returns

MPI status object

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Recv() [6/10]

MPI_Status monolish::mpi::comm::Recv	(	size_t	val,
		int	src,
		int	tag
	)		const

MPI_Recv for scalar. Performs a blocking recv.

Parameters

val	scalar value
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [7/10]

MPI_Status monolish::mpi::comm::Recv	(	std::vector< double > &	vec,
		int	src,
		int	tag
	)		const

MPI_Recv for std::vector. Performs a blocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [8/10]

MPI_Status monolish::mpi::comm::Recv	(	std::vector< float > &	vec,
		int	src,
		int	tag
	)		const

MPI_Recv for std::vector. Performs a blocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [9/10]

MPI_Status monolish::mpi::comm::Recv	(	std::vector< int > &	vec,
		int	src,
		int	tag
	)		const

MPI_Recv for std::vector. Performs a blocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Returns

MPI status object

Recv() [10/10]

MPI_Status monolish::mpi::comm::Recv	(	std::vector< size_t > &	vec,
		int	src,
		int	tag
	)		const

MPI_Recv for std::vector. Performs a blocking recv.

Parameters

vec	std::vector (size N)
src	rank of source
tag	message tag

Returns

MPI status object

Scatter() [1/6]

void monolish::mpi::comm::Scatter	(	monolish::vector< double > &	sendvec,
		monolish::vector< double > &	recvvec,
		int	root
	)		const

MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, monolish vector (size N)
recvvec	recv data, monolish vector (size N / # of procs)
root	root rank number

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Scatter() [2/6]

void monolish::mpi::comm::Scatter	(	monolish::vector< float > &	sendvec,
		monolish::vector< float > &	recvvec,
		int	root
	)		const

MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, monolish vector (size N)
recvvec	recv data, monolish vector (size N / # of procs)
root	root rank number

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Scatter() [3/6]

void monolish::mpi::comm::Scatter	(	std::vector< double > &	sendvec,
		std::vector< double > &	recvvec,
		int	root
	)		const

MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std::vector (size N)
recvvec	recv data, std::vector (size N / # of procs)
root	root rank number

Scatter() [4/6]

void monolish::mpi::comm::Scatter	(	std::vector< float > &	sendvec,
		std::vector< float > &	recvvec,
		int	root
	)		const

MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std::vector (size N)
recvvec	recv data, std::vector (size N / # of procs)
root	root rank number

Scatter() [5/6]

void monolish::mpi::comm::Scatter	(	std::vector< int > &	sendvec,
		std::vector< int > &	recvvec,
		int	root
	)		const

MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std::vector (size N)
recvvec	recv data, std::vector (size N / # of procs)
root	root rank number

Scatter() [6/6]

void monolish::mpi::comm::Scatter	(	std::vector< size_t > &	sendvec,
		std::vector< size_t > &	recvvec,
		int	root
	)		const

MPI_Scatter, Sends data from one task to all tasks. The data is evenly divided and transmitted to each process.

Parameters

sendvec	send data, std::vector (size N)
recvvec	recv data, std::vector (size N / # of procs)
root	root rank number

Send() [1/10]

void monolish::mpi::comm::Send	(	double	val,
		int	dst,
		int	tag
	)		const

MPI_Send for scalar. Performs a blocking send.

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Send() [2/10]

void monolish::mpi::comm::Send	(	float	val,
		int	dst,
		int	tag
	)		const

MPI_Send for scalar. Performs a blocking send.

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Send() [3/10]

void monolish::mpi::comm::Send	(	int	val,
		int	dst,
		int	tag
	)		const

MPI_Send for scalar. Performs a blocking send.

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Send() [4/10]

void monolish::mpi::comm::Send	(	monolish::vector< double > &	vec,
		int	dst,
		int	tag
	)		const

MPI_Send for monolish::vector. Performs a blocking send.

Parameters

vec	monolish::vector (size N)
dst	rank of destination
tag	message tag

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Send() [5/10]

void monolish::mpi::comm::Send	(	monolish::vector< float > &	vec,
		int	dst,
		int	tag
	)		const

MPI_Send for monolish::vector. Performs a blocking send.

Parameters

vec	monolish::vector (size N)
dst	rank of destination
tag	message tag

Warning

MPI functions do not support GPUs. The user needs to send and receive data to and from the GPU before and after the MPI function.

Send() [6/10]

void monolish::mpi::comm::Send	(	size_t	val,
		int	dst,
		int	tag
	)		const

MPI_Send for scalar. Performs a blocking send.

Parameters

val	scalar value
dst	rank of destination
tag	message tag

Send() [7/10]

void monolish::mpi::comm::Send	(	std::vector< double > &	vec,
		int	dst,
		int	tag
	)		const

MPI_Send for std::vector. Performs a blocking send.

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Send() [8/10]

void monolish::mpi::comm::Send	(	std::vector< float > &	vec,
		int	dst,
		int	tag
	)		const

MPI_Send for std::vector. Performs a blocking send.

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Send() [9/10]

void monolish::mpi::comm::Send	(	std::vector< int > &	vec,
		int	dst,
		int	tag
	)		const

MPI_Send for std::vector. Performs a blocking send.

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

Send() [10/10]

void monolish::mpi::comm::Send	(	std::vector< size_t > &	vec,
		int	dst,
		int	tag
	)		const

MPI_Send for std::vector. Performs a blocking send.

Parameters

vec	std::vector (size N)
dst	rank of destination
tag	message tag

set_comm()

void monolish::mpi::comm::set_comm

(

MPI_Comm

external_comm

)

set communicator

Waitall()

void monolish::mpi::comm::Waitall

(

)

Waits for all communications to complete.

Member Data Documentation

my_comm

MPI_Comm monolish::mpi::comm::my_comm = 0

private

MPI communicator, MPI_COMM_WORLD.

Definition at line 27 of file monolish_mpi_core.hpp.

requests

std::vector<MPI_Request> monolish::mpi::comm::requests

private

Definition at line 31 of file monolish_mpi_core.hpp.

---

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

• monolish_mpi_core.hpp