Composable Kernel: ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument Struct Reference

Argument Struct Reference

Argument Struct Reference#

Composable Kernel: ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument Struct Reference

Public Member Functions | Public Attributes | List of all members

ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument Struct Reference

#include <device_batchnorm_forward_impl.hpp>

Inheritance diagram for ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument:

Public Member Functions
	Argument (const std::array< index_t, Rank > xyLengths, const std::array< index_t, Rank > xStrides, const std::array< index_t, Rank > yStrides, const std::array< int, NumBatchNormReduceDim > reduceDims, const std::array< index_t, Rank - NumBatchNormReduceDim > bnScaleBiasMeanVarLengths, const std::array< index_t, Rank - NumBatchNormReduceDim > bnScaleStrides, const std::array< index_t, Rank - NumBatchNormReduceDim > bnBiasStrides, const std::array< index_t, Rank - NumBatchNormReduceDim > bnMeanVarStrides, const XDataType p_x, const ScaleDataType p_scale, const BiasDataType p_bias, const YElementwiseOp y_elementwise_op, double epsilon, YDataType p_y, MeanVarDataType resultSaveMean, MeanVarDataType resultSaveInvVariance, double averageFactor, MeanVarDataType resultRunningMean, MeanVarDataType resultRunningVariance)
	Argument (const std::array< index_t, Rank > xyLengths, const std::array< index_t, Rank > xStrides, const std::array< index_t, Rank > yStrides, const std::array< int, NumBatchNormReduceDim > reduceDims, const std::array< index_t, Rank - NumBatchNormReduceDim > bnScaleBiasMeanVarLengths, const std::array< index_t, Rank - NumBatchNormReduceDim > bnScaleStrides, const std::array< index_t, Rank - NumBatchNormReduceDim > bnBiasStrides, const std::array< index_t, Rank - NumBatchNormReduceDim > bnMeanVarStrides, const XDataType p_x, const ScaleDataType p_scale, const BiasDataType p_bias, const YElementwiseOp y_elementwise_op, double epsilon, YDataType p_y, MeanVarDataType resultSaveMean, MeanVarDataType resultSaveInvVariance, double averageFactor, MeanVarDataType resultRunningMean, MeanVarDataType resultRunningVariance)
Public Member Functions inherited from ck::tensor_operation::device::BaseArgument
	BaseArgument ()=default
	BaseArgument (const BaseArgument &)=default
BaseArgument &	operator= (const BaseArgument &)=default
virtual	~BaseArgument ()

Public Attributes
AccDataType	epsilon_
AccDataType	averageFactor_
bool	updateMovingAverage_
bool	saveMeanInvVariance_
std::array< index_t, Rank >	xyLengths_
std::array< index_t, Rank >	xStrides_
std::array< index_t, Rank >	yStrides_
std::array< index_t, Rank - NumBatchNormReduceDim >	bnScaleBiasMeanVarLengths_
std::array< index_t, Rank - NumBatchNormReduceDim >	bnScaleStrides_
std::array< index_t, Rank - NumBatchNormReduceDim >	bnBiasStrides_
std::array< index_t, Rank - NumBatchNormReduceDim >	bnMeanVarStrides_
const XDataType *	p_x_
const ScaleDataType *	p_scale_
const BiasDataType *	p_bias_
const YElementwiseOp	y_elementwise_op_
YDataType *	p_y_
MeanVarDataType *	resultSaveMean_
MeanVarDataType *	resultSaveInvVariance_
MeanVarDataType *	resultRunningMean_
MeanVarDataType *	resultRunningVariance_
long_index_t	invariant_length_
long_index_t	reduce_length_
int	blkGroupSize_
int	numBlockTileIteration_
size_t	gridSize_
XYGridDesc_M_K	x_grid_desc_m_k_
XYGridDesc_M_K	y_grid_desc_m_k_
ScaleBiasMeanVarGridDesc_M	scale_grid_desc_m_
ScaleBiasMeanVarGridDesc_M	bias_grid_desc_m_
ScaleBiasMeanVarGridDesc_M	mean_var_grid_desc_m_
void *	workspace_mean_
void *	workspace_variance_
void *	workspace_count_
void *	control_
Public Attributes inherited from ck::tensor_operation::device::BaseArgument
void *	p_workspace_ = nullptr

Constructor & Destructor Documentation

◆ Argument() [1/2]

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::Argument	(	const std::array< index_t, Rank >	xyLengths,
		const std::array< index_t, Rank >	xStrides,
		const std::array< index_t, Rank >	yStrides,
		const std::array< int, NumBatchNormReduceDim >	reduceDims,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnScaleBiasMeanVarLengths,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnScaleStrides,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnBiasStrides,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnMeanVarStrides,
		const XDataType *	p_x,
		const ScaleDataType *	p_scale,
		const BiasDataType *	p_bias,
		const YElementwiseOp	y_elementwise_op,
		double	epsilon,
		YDataType *	p_y,
		MeanVarDataType *	resultSaveMean,
		MeanVarDataType *	resultSaveInvVariance,
		double	averageFactor,
		MeanVarDataType *	resultRunningMean,
		MeanVarDataType *	resultRunningVariance )

inline

◆ Argument() [2/2]

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::Argument	(	const std::array< index_t, Rank >	xyLengths,
		const std::array< index_t, Rank >	xStrides,
		const std::array< index_t, Rank >	yStrides,
		const std::array< int, NumBatchNormReduceDim >	reduceDims,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnScaleBiasMeanVarLengths,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnScaleStrides,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnBiasStrides,
		const std::array< index_t, Rank - NumBatchNormReduceDim >	bnMeanVarStrides,
		const XDataType *	p_x,
		const ScaleDataType *	p_scale,
		const BiasDataType *	p_bias,
		const YElementwiseOp	y_elementwise_op,
		double	epsilon,
		YDataType *	p_y,
		MeanVarDataType *	resultSaveMean,
		MeanVarDataType *	resultSaveInvVariance,
		double	averageFactor,
		MeanVarDataType *	resultRunningMean,
		MeanVarDataType *	resultRunningVariance )

inline

Member Data Documentation

◆ averageFactor_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

AccDataType ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::averageFactor_

◆ bias_grid_desc_m_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

ScaleBiasMeanVarGridDesc_M ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::bias_grid_desc_m_

◆ blkGroupSize_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

int ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::blkGroupSize_

◆ bnBiasStrides_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank - NumBatchNormReduceDim > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::bnBiasStrides_

◆ bnMeanVarStrides_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank - NumBatchNormReduceDim > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::bnMeanVarStrides_

◆ bnScaleBiasMeanVarLengths_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank - NumBatchNormReduceDim > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::bnScaleBiasMeanVarLengths_

◆ bnScaleStrides_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank - NumBatchNormReduceDim > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::bnScaleStrides_

◆ control_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

void* ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::control_

◆ epsilon_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

AccDataType ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::epsilon_

◆ gridSize_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

size_t ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::gridSize_

◆ invariant_length_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

long_index_t ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::invariant_length_

◆ mean_var_grid_desc_m_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

ScaleBiasMeanVarGridDesc_M ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::mean_var_grid_desc_m_

◆ numBlockTileIteration_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

int ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::numBlockTileIteration_

◆ p_bias_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

const BiasDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::p_bias_

◆ p_scale_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

const ScaleDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::p_scale_

◆ p_x_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

const XDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::p_x_

◆ p_y_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

YDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::p_y_

◆ reduce_length_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

long_index_t ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::reduce_length_

◆ resultRunningMean_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

MeanVarDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::resultRunningMean_

◆ resultRunningVariance_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

MeanVarDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::resultRunningVariance_

◆ resultSaveInvVariance_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

MeanVarDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::resultSaveInvVariance_

◆ resultSaveMean_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

MeanVarDataType * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::resultSaveMean_

◆ saveMeanInvVariance_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

bool ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::saveMeanInvVariance_

◆ scale_grid_desc_m_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

ScaleBiasMeanVarGridDesc_M ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::scale_grid_desc_m_

◆ updateMovingAverage_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

bool ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::updateMovingAverage_

◆ workspace_count_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

void * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::workspace_count_

◆ workspace_mean_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

void * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::workspace_mean_

◆ workspace_variance_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

void * ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::workspace_variance_

◆ x_grid_desc_m_k_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

XYGridDesc_M_K ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::x_grid_desc_m_k_

◆ xStrides_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::xStrides_

◆ xyLengths_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::xyLengths_

◆ y_elementwise_op_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

const YElementwiseOp ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::y_elementwise_op_

◆ y_grid_desc_m_k_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

XYGridDesc_M_K ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::y_grid_desc_m_k_

◆ yStrides_

template<typename XDataType, typename YDataType, typename AccDataType, typename ScaleDataType, typename BiasDataType, typename MeanVarDataType, typename YElementwiseOp, index_t Rank, index_t NumBatchNormReduceDim, bool UseMultiblockInK, index_t BlockSize, index_t MThreadClusterSize, index_t KThreadClusterSize, index_t MThreadSliceSize, index_t KThreadSliceSize, index_t XSrcYDstVectorDim, index_t XSrcVectorSize, index_t YDstVectorSize, index_t ScaleSrcVectorSize, index_t BiasSrcVectorSize, index_t MeanVarSrcDstVectorSize>

std::array< index_t, Rank > ck::tensor_operation::device::DeviceBatchNormFwdImpl< XDataType, YDataType, AccDataType, ScaleDataType, BiasDataType, MeanVarDataType, YElementwiseOp, Rank, NumBatchNormReduceDim, UseMultiblockInK, BlockSize, MThreadClusterSize, KThreadClusterSize, MThreadSliceSize, KThreadSliceSize, XSrcYDstVectorDim, XSrcVectorSize, YDstVectorSize, ScaleSrcVectorSize, BiasSrcVectorSize, MeanVarSrcDstVectorSize >::Argument::yStrides_

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