ibmppl::linear_task_partitioner< T > Class Template Reference

implement block-based linear partitioning for container class T More...

#include <for_each.h>

Public Types
typedef T::iterator	iterator
typedef T::const_iterator	const_iterator

Public Member Functions
	linear_task_partitioner (T &data, PARTITION_TYPE ptype, size_t chunk_size)
	linear_task_partitioner ()
size_t	num_partitions ()
iterator	partition_begin (task_id chunk_id)
iterator	partition_end (task_id chunk_id)

Protected Member Functions
void	setData (T *data)
void	init (PARTITION_TYPE ptype, size_t chunk_size)

Protected Attributes
size_t	_chunk_size
size_t	_num_chunks
size_t	_total_size
PARTITION_TYPE	_ptype
T *	_data

Detailed Description

template<class T>
class ibmppl::linear_task_partitioner< T >

implement block-based linear partitioning for container class T

class <T> implement the following interface: T::size(), T::begin(), T::end(), T::iterator, and T::const_iterator

ptype:

• STATIC_PARTITION: T is divided into <num_threads> # of tasks equally
• BLOCK_PARTITION: each task is of size <n> elements

chunk_size:

• for STATIC_PARTITION: chunk_size is minimal chunk size for task (if not enough tasks, # of tasks will be less than # threads
• for BLOCK_PARTITION: chunk_size is the task chunk size

Constructor & Destructor Documentation

template<class T>

ibmppl::linear_task_partitioner< T >::linear_task_partitioner ( T &  data, PARTITION_TYPE  ptype, size_t  chunk_size  )

inline

constructor: building a linear task partitioner container <data>

template<class T>

ibmppl::linear_task_partitioner< T >::linear_task_partitioner ( )

inline

constructor: dummy (not doing anything)

Member Function Documentation

template<class T>

size_t ibmppl::linear_task_partitioner< T >::num_partitions ( )

inline

returns number of partitions

template<class T>

iterator ibmppl::linear_task_partitioner< T >::partition_begin ( task_id  chunk_id )

inline

returns iterator to the beginning of a partition specified by chunk_id

template<class T>

iterator ibmppl::linear_task_partitioner< T >::partition_end ( task_id  chunk_id )

inline

returns iterator to the end of a partition specified by chunk_id

---

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

• /home/yxia/Work/ibmppl.gsa/runtime/for_each.h