CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers
Public Member Functions | Public Attributes | List of all members
cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params Struct Reference

Params structure.

#include <reduce_split_k.h>

Collaboration diagram for cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params:
Collaboration graph
[legend]

Public Member Functions

CUTLASS_HOST_DEVICE Params ()
 
CUTLASS_HOST_DEVICE Params (MatrixCoord problem_size_, int partitions_, size_t partition_stride_, WorkspaceTensorRef workspace_, OutputTensorRef destination_, OutputTensorRef source_, typename OutputOp::Params output_=typename OutputOp::Params(), typename ReductionOp::Params reduction_=typename ReductionOp::Params())
 

Public Attributes

MatrixCoord problem_size
 
int partitions
 
size_t partition_stride
 
WorkspaceTensorRef workspace
 
OutputTensorRef destination
 
OutputTensorRef source
 
OutputOp::Params output
 
ReductionOp::Params reduction
 

Constructor & Destructor Documentation

template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
CUTLASS_HOST_DEVICE cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::Params ( )
inline
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
CUTLASS_HOST_DEVICE cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::Params ( MatrixCoord  problem_size_,
int  partitions_,
size_t  partition_stride_,
WorkspaceTensorRef  workspace_,
OutputTensorRef  destination_,
OutputTensorRef  source_,
typename OutputOp::Params  output_ = typename OutputOp::Params(),
typename ReductionOp::Params  reduction_ = typename ReductionOp::Params() 
)
inline

Member Data Documentation

template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
OutputTensorRef cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::destination
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
OutputOp::Params cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::output
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
size_t cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::partition_stride
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
int cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::partitions
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
MatrixCoord cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::problem_size
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
ReductionOp::Params cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::reduction
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
OutputTensorRef cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::source
template<typename Shape_ , typename OutputOp_ , typename ReductionOp_ , int PartitionsPerStage = 4>
WorkspaceTensorRef cutlass::reduction::kernel::ReduceSplitK< Shape_, OutputOp_, ReductionOp_, PartitionsPerStage >::Params::workspace

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