sc::MemoryGrpRegion Class Reference

The MemoryGrpRegion is a MemoryGrp proxy to a region of a MemoryGrp. More...

#include <util/group/memregion.h>

Inheritance diagram for sc::MemoryGrpRegion:

List of all members.

Public Member Functions

MemoryGrpRegion (const Ref< MemoryGrp > &mem, size_t mem_start, size_t max_size)

MemoryGrpRegion is defined by the host MemoryGrp and position and size of the local segment within the local memory of the host.

MemoryGrpRegion (const Ref< MemoryGrp > &mem, size_t max_size)

same as above, except mem_start is determined automatically

void set_localsize (size_t)

Set the size of locally held memory.

void * localdata ()

Returns a pointer to the local data.

void * obtain_readwrite (distsize_t offset, int size)

Only one thread can have an unreleased obtain_readwrite at a time.

void * obtain_readonly (distsize_t offset, int size)

This gives read access to the memory location. No locking is done.

void * obtain_writeonly (distsize_t offset, int size)

This gives write access to the memory location. No locking is done.

void release_readonly (void *data, distsize_t offset, int size)

This is called when read access is no longer needed.

void release_writeonly (void *data, distsize_t offset, int size)

This is called when write access is no longer needed.

void release_readwrite (void *data, distsize_t offset, int size)

This is called when read/write access is no longer needed.

void sum_reduction_on_node (double *data, size_t doffset, int dsize, int node=-1)

Perform a sum reduction on double data localized to a single node.

void activate ()

Activate is called before the memory is to be used.

void deactivate ()

Deactivate is called after the memory has been used.

void sync ()

Synchronizes all the nodes.

void catchup ()

Processes outstanding requests.

void * malloc_local (size_t nbytes)

Allocate data that will be accessed locally only.

void free_local (void *data)

Free data that was allocated with malloc_local.

Ref< MemoryGrp > clone ()

reimplementation of MemoryGrp::clone(). always throws ProgrammingError -- this type of MemoryGrp is not clonable

void print (std::ostream &o=ExEnv::out0()) const

Prints out information about the object.

Classes

class HostToRegionsMap

this maps MemoryGrp to the ordered list of its allocated regions, in order of increasing starts

class LocalRegion

LocalRegion is an interval [start, start+size).

Detailed Description

The MemoryGrpRegion is a MemoryGrp proxy to a region of a MemoryGrp.

It is assumed that the host MemoryGrp is not resized during lifetime of all MemoryGrpRegion objects bound to it.

Each region has a static local offset and size. set_localsize() then cannot ask for more than the maximum size, but can be called multiple times.

Member Function Documentation

void sc::MemoryGrpRegion::set_localsize ( size_t ) [virtual]

Set the size of locally held memory.

When memory is accessed using a global offset counting starts at node 0 and proceeds up to node n() - 1.

Implements sc::MemoryGrp.

void* sc::MemoryGrpRegion::obtain_readwrite	(	distsize_t	offset,
		int	size
	)			`[virtual]`

Only one thread can have an unreleased obtain_readwrite at a time.

The actual memory region locked can be larger than that requested. If the memory region is already locked this will block. For this reason, data should be held as read/write for as short a time as possible.

Implements sc::MemoryGrp.

void sc::MemoryGrpRegion::release_readwrite	(	void *	data,
		distsize_t	offset,
		int	size
	)			`[virtual]`

This is called when read/write access is no longer needed.

The memory will be unlocked.

Implements sc::MemoryGrp.

void sc::MemoryGrpRegion::sum_reduction_on_node	(	double *	data,
		size_t	doffset,
		int	dsize,
		int	node = `-1`
	)			`[virtual]`

Perform a sum reduction on double data localized to a single node.

Parameters:

	data	the contribution to sum into the global array.
	doffset	the local offset on the node in terms of doubles.
	dsize	the size in terms of doubles.
	node	the node holding the data. The default is the local node.

Reimplemented from sc::MemoryGrp.

void sc::MemoryGrpRegion::sync ( ) [virtual]

Synchronizes all the nodes.

This is useful after remote memory writes to be certain that all of the writes have completed and the data can be accessed locally, for example.

Implements sc::MemoryGrp.

void sc::MemoryGrpRegion::catchup ( ) [virtual]

Processes outstanding requests.

Some memory group implementations don't have access to real shared memory or even active messages. Instead, requests are processed whenever certain memory group routines are called. This can cause large latencies and buffer overflows. If this is a problem, then the catchup member can be called to process all outstanding requests.

Reimplemented from sc::MemoryGrp.

void* sc::MemoryGrpRegion::malloc_local ( size_t nbyte ) [virtual]

Allocate data that will be accessed locally only.

Using this for data that will be used for global operations can improve efficiency. Data allocated in this way must be freed with free_local_double.

Reimplemented from sc::MemoryGrp.

void sc::MemoryGrpRegion::free_local ( void * data ) [virtual]

Free data that was allocated with malloc_local.

Reimplemented from sc::MemoryGrp.

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

src/lib/util/group/memregion.h


Public Member Functions
	MemoryGrpRegion (const Ref< MemoryGrp > &mem, size_t mem_start, size_t max_size)
	MemoryGrpRegion is defined by the host MemoryGrp and position and size of the local segment within the local memory of the host.
	MemoryGrpRegion (const Ref< MemoryGrp > &mem, size_t max_size)
	same as above, except mem_start is determined automatically
void	set_localsize (size_t)
	Set the size of locally held memory.
void *	localdata ()
	Returns a pointer to the local data.
void *	obtain_readwrite (distsize_t offset, int size)
	Only one thread can have an unreleased obtain_readwrite at a time.
void *	obtain_readonly (distsize_t offset, int size)
	This gives read access to the memory location. No locking is done.
void *	obtain_writeonly (distsize_t offset, int size)
	This gives write access to the memory location. No locking is done.
void	release_readonly (void *data, distsize_t offset, int size)
	This is called when read access is no longer needed.
void	release_writeonly (void *data, distsize_t offset, int size)
	This is called when write access is no longer needed.
void	release_readwrite (void *data, distsize_t offset, int size)
	This is called when read/write access is no longer needed.
void	sum_reduction_on_node (double *data, size_t doffset, int dsize, int node=-1)
	Perform a sum reduction on double data localized to a single node.
void	activate ()
	Activate is called before the memory is to be used.
void	deactivate ()
	Deactivate is called after the memory has been used.
void	sync ()
	Synchronizes all the nodes.
void	catchup ()
	Processes outstanding requests.
void *	malloc_local (size_t nbytes)
	Allocate data that will be accessed locally only.
void	free_local (void *data)
	Free data that was allocated with malloc_local.
Ref< MemoryGrp >	clone ()
	reimplementation of MemoryGrp::clone(). always throws ProgrammingError -- this type of MemoryGrp is not clonable
void	print (std::ostream &o=ExEnv::out0()) const
	Prints out information about the object.
Classes
class	HostToRegionsMap
	this maps MemoryGrp to the ordered list of its allocated regions, in order of increasing starts
class	LocalRegion
	LocalRegion is an interval [start, start+size).