< Binary reduction functor type having member T operator()(const T &a, const T &b) More...

Detailed Description

template<typename InputIteratorT, typename OutputIteratorT, typename OffsetT, typename ReductionOpT>
struct cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >

< Binary reduction functor type having member T operator()(const T &a, const T &b)

Utility class for dispatching the appropriately-tuned kernels for device-wide reduction

Definition at line 359 of file dispatch_reduce.cuh.

Inheritance diagram for cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >:

Public Types
typedef If<(Equals< typenamestd::iterator_traits< OutputIteratorT >::value_type, void >::VALUE), typenamestd::iterator_traits< InputIteratorT >::value_type, typenamestd::iterator_traits< OutputIteratorT >::value_type >::Type	OutputT

Public Types inherited from cub::DeviceReducePolicy< If<(Equals< std::iterator_traits< OutputIteratorT >::value_type, void >::VALUE), std::iterator_traits< InputIteratorT >::value_type, std::iterator_traits< OutputIteratorT >::value_type >::Type, OffsetT, ReductionOpT >
typedef Policy600	MaxPolicy
	MaxPolicy.

Public Member Functions
CUB_RUNTIME_FUNCTION __forceinline__	DispatchReduce (void *d_temp_storage, size_t &temp_storage_bytes, InputIteratorT d_in, OutputIteratorT d_out, OffsetT num_items, ReductionOpT reduction_op, OutputT init, cudaStream_t stream, bool debug_synchronous, int ptx_version)
	Constructor.

template<typename ActivePolicyT , typename SingleTileKernelT >
CUB_RUNTIME_FUNCTION __forceinline__ cudaError_t	InvokeSingleTile (SingleTileKernelT single_tile_kernel)
	Invoke a single block block to reduce in-core.

template<typename ActivePolicyT , typename ReduceKernelT , typename SingleTileKernelT >
CUB_RUNTIME_FUNCTION __forceinline__ cudaError_t	InvokePasses (ReduceKernelT reduce_kernel, SingleTileKernelT single_tile_kernel)
	Invoke two-passes to reduce.

template<typename ActivePolicyT >
CUB_RUNTIME_FUNCTION __forceinline__ cudaError_t	Invoke ()
	Invocation.

Static Public Member Functions
CUB_RUNTIME_FUNCTION static __forceinline__ cudaError_t	Dispatch (void *d_temp_storage, size_t &temp_storage_bytes, InputIteratorT d_in, OutputIteratorT d_out, OffsetT num_items, ReductionOpT reduction_op, OutputT init, cudaStream_t stream, bool debug_synchronous)

Data Fields
void *	d_temp_storage
	[in] Device-accessible allocation of temporary storage. When NULL, the required allocation size is written to `temp_storage_bytes` and no work is done.

size_t &	temp_storage_bytes
	[in,out] Reference to size in bytes of `d_temp_storage` allocation

InputIteratorT	d_in
	[in] Pointer to the input sequence of data items

OutputIteratorT	d_out
	[out] Pointer to the output aggregate

OffsetT	num_items
	[in] Total number of input items (i.e., length of `d_in`)

ReductionOpT	reduction_op
	[in] Binary reduction functor

OutputT	init
	[in] The initial value of the reduction

cudaStream_t	stream
	[in] CUDA stream to launch kernels within. Default is stream₀.

bool	debug_synchronous
	[in] Whether or not to synchronize the stream after every kernel launch to check for errors. Also causes launch configurations to be printed to the console. Default is `false`.

int	ptx_version
	[in] PTX version

Member Typedef Documentation

◆ OutputT

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

typedef If<(Equals<typenamestd::iterator_traits<OutputIteratorT>::value_type,void>::VALUE),typenamestd::iterator_traits<InputIteratorT>::value_type,typenamestd::iterator_traits<OutputIteratorT>::value_type>::Type cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::OutputT

Definition at line 374 of file dispatch_reduce.cuh.

Constructor & Destructor Documentation

◆ DispatchReduce()

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

CUB_RUNTIME_FUNCTION __forceinline__ cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::DispatchReduce	(	void *	d_temp_storage,
		size_t &	temp_storage_bytes,
		InputIteratorT	d_in,
		OutputIteratorT	d_out,
		OffsetT	num_items,
		ReductionOpT	reduction_op,
		OutputT	init,
		cudaStream_t	stream,
		bool	debug_synchronous,
		int	ptx_version
	)

inline

Constructor.

Definition at line 398 of file dispatch_reduce.cuh.

Member Function Documentation

◆ Dispatch()

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

CUB_RUNTIME_FUNCTION static __forceinline__ cudaError_t cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::Dispatch	(	void *	d_temp_storage,
		size_t &	temp_storage_bytes,
		InputIteratorT	d_in,
		OutputIteratorT	d_out,
		OffsetT	num_items,
		ReductionOpT	reduction_op,
		OutputT	init,
		cudaStream_t	stream,
		bool	debug_synchronous
	)

inlinestatic

Internal dispatch routine for computing a device-wide reduction

Parameters

[in]	d_temp_storage	Device-accessible allocation of temporary storage. When NULL, the required allocation size is written to `temp_storage_bytes` and no work is done.
[in,out]	temp_storage_bytes	Reference to size in bytes of `d_temp_storage` allocation
[in]	d_in	Pointer to the input sequence of data items
[out]	d_out	Pointer to the output aggregate
[in]	num_items	Total number of input items (i.e., length of `d_in`)
[in]	reduction_op	Binary reduction functor
[in]	init	The initial value of the reduction
[in]	stream	[optional] CUDA stream to launch kernels within. Default is stream₀.
[in]	debug_synchronous	[optional] Whether or not to synchronize the stream after every kernel launch to check for errors. Also causes launch configurations to be printed to the console. Default is `false`.

Definition at line 631 of file dispatch_reduce.cuh.

◆ Invoke()

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

template<typename ActivePolicyT >

CUB_RUNTIME_FUNCTION __forceinline__ cudaError_t cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::Invoke ( )

inline

Invocation.

Definition at line 601 of file dispatch_reduce.cuh.

◆ InvokePasses()

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

template<typename ActivePolicyT , typename ReduceKernelT , typename SingleTileKernelT >

CUB_RUNTIME_FUNCTION __forceinline__ cudaError_t cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::InvokePasses	(	ReduceKernelT	reduce_kernel,
		SingleTileKernelT	single_tile_kernel
	)

inline

Invoke two-passes to reduce.

< Function type of cub::DeviceReduceSingleTileKernel

Parameters

[in]	reduce_kernel	Kernel function pointer to parameterization of cub::DeviceReduceKernel
[in]	single_tile_kernel	Kernel function pointer to parameterization of cub::DeviceReduceSingleTileKernel

Definition at line 489 of file dispatch_reduce.cuh.

◆ InvokeSingleTile()

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

template<typename ActivePolicyT , typename SingleTileKernelT >

CUB_RUNTIME_FUNCTION __forceinline__ cudaError_t cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::InvokeSingleTile ( SingleTileKernelT single_tile_kernel )

inline

Invoke a single block block to reduce in-core.

< Function type of cub::DeviceReduceSingleTileKernel

Parameters

[in] single_tile_kernel Kernel function pointer to parameterization of cub::DeviceReduceSingleTileKernel

Definition at line 432 of file dispatch_reduce.cuh.

Field Documentation

◆ d_in

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

InputIteratorT cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::d_in

[in] Pointer to the input sequence of data items

Definition at line 383 of file dispatch_reduce.cuh.

◆ d_out

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

OutputIteratorT cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::d_out

[out] Pointer to the output aggregate

Definition at line 384 of file dispatch_reduce.cuh.

◆ d_temp_storage

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

void* cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::d_temp_storage

[in] Device-accessible allocation of temporary storage. When NULL, the required allocation size is written to temp_storage_bytes and no work is done.

Definition at line 381 of file dispatch_reduce.cuh.

◆ debug_synchronous

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

bool cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::debug_synchronous

[in] Whether or not to synchronize the stream after every kernel launch to check for errors. Also causes launch configurations to be printed to the console. Default is false.

Definition at line 389 of file dispatch_reduce.cuh.

◆ init

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

OutputT cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::init

[in] The initial value of the reduction

Definition at line 387 of file dispatch_reduce.cuh.

◆ num_items

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

OffsetT cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::num_items

[in] Total number of input items (i.e., length of d_in)

Definition at line 385 of file dispatch_reduce.cuh.

◆ ptx_version

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

int cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::ptx_version

[in] PTX version

Definition at line 390 of file dispatch_reduce.cuh.

◆ reduction_op

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

ReductionOpT cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::reduction_op

[in] Binary reduction functor

Definition at line 386 of file dispatch_reduce.cuh.

◆ stream

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

cudaStream_t cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::stream

[in] CUDA stream to launch kernels within. Default is stream₀.

Definition at line 388 of file dispatch_reduce.cuh.

◆ temp_storage_bytes

template<typename InputIteratorT , typename OutputIteratorT , typename OffsetT , typename ReductionOpT >

size_t& cub::DispatchReduce< InputIteratorT, OutputIteratorT, OffsetT, ReductionOpT >::temp_storage_bytes

[in,out] Reference to size in bytes of d_temp_storage allocation

Definition at line 382 of file dispatch_reduce.cuh.

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

openfpm_data/src/util/cuda/cub_old/device/dispatch/dispatch_reduce.cuh

Detailed Description

Public Types

Public Member Functions

Static Public Member Functions

Data Fields

Member Typedef Documentation

◆ OutputT

Constructor & Destructor Documentation

◆ DispatchReduce()

Member Function Documentation

◆ Dispatch()

◆ Invoke()

◆ InvokePasses()

◆ InvokeSingleTile()

Field Documentation

◆ d_in

◆ d_out

◆ d_temp_storage

◆ debug_synchronous

◆ init

◆ num_items

◆ ptx_version

◆ reduction_op

◆ stream

◆ temp_storage_bytes