device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp File Reference

device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp File Reference#

Composable Kernel: device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp File Reference
Classes | Namespaces | Macros
device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp File Reference
#include <iostream>
#include <sstream>
#include "ck/library/utility/numeric.hpp"
#include "ck/utility/common_header.hpp"
#include "ck/utility/env.hpp"
#include "ck/tensor_description/tensor_descriptor.hpp"
#include "ck/tensor_description/tensor_descriptor_helper.hpp"
#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
#include "ck/tensor_operation/gpu/device/device_grouped_conv_bwd_data_multiple_d.hpp"
#include "ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp"
#include "ck/tensor_operation/operator_transform/transform_conv_bwd_data_to_gemm_v1.hpp"
#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
#include "ck/tensor_operation/operator_transform/transform_conv_ngchw_to_nhwgc.hpp"
#include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
#include "ck/tensor_operation/gpu/grid/gridwise_elementwise_2d.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_grouped_conv_utils.hpp"
#include "ck/host_utility/device_prop.hpp"
#include "ck/host_utility/kernel_launch.hpp"
#include "ck/host_utility/flush_cache.hpp"
#include "ck/host_utility/io.hpp"
#include "ck/tensor_operation/gpu/grid/block_to_ctile_map.hpp"

Go to the source code of this file.

Classes

struct  ck::tensor_operation::device::DeviceGroupedConvBwdDataMultipleD_Xdl_CShuffle_v1< NDimSpatial, ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOp, BElementwiseOp, CDEElementwiseOp, ConvBackwardDataSpecialization, DoPadGemmM, DoPadGemmN, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, LoopSched, AComputeType, BComputeType, MaxTransposeTransferInScalarPerVector, MaxTransposeTransferOutScalarPerVector >
struct  ck::tensor_operation::device::DeviceGroupedConvBwdDataMultipleD_Xdl_CShuffle_v1< NDimSpatial, ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOp, BElementwiseOp, CDEElementwiseOp, ConvBackwardDataSpecialization, DoPadGemmM, DoPadGemmN, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, LoopSched, AComputeType, BComputeType, MaxTransposeTransferInScalarPerVector, MaxTransposeTransferOutScalarPerVector >::GemmArgs
struct  ck::tensor_operation::device::DeviceGroupedConvBwdDataMultipleD_Xdl_CShuffle_v1< NDimSpatial, ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOp, BElementwiseOp, CDEElementwiseOp, ConvBackwardDataSpecialization, DoPadGemmM, DoPadGemmN, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, LoopSched, AComputeType, BComputeType, MaxTransposeTransferInScalarPerVector, MaxTransposeTransferOutScalarPerVector >::Argument
struct  ck::tensor_operation::device::DeviceGroupedConvBwdDataMultipleD_Xdl_CShuffle_v1< NDimSpatial, ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType, AElementwiseOp, BElementwiseOp, CDEElementwiseOp, ConvBackwardDataSpecialization, DoPadGemmM, DoPadGemmN, NumGemmKPrefetchStage, BlockSize, MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave, ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, BBlockLdsExtraN, CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, CDEBlockTransferScalarPerVector_NPerBlock, LoopSched, AComputeType, BComputeType, MaxTransposeTransferInScalarPerVector, MaxTransposeTransferOutScalarPerVector >::Invoker

Namespaces

namespace  ck
namespace  ck::tensor_operation
namespace  ck::tensor_operation::device

Macros

#define GridwiseGemmMultiDTemplateParams
#define GridwiseGemmCTransposeTemplateParameters

Macro Definition Documentation

◆ GridwiseGemmCTransposeTemplateParameters

#define GridwiseGemmCTransposeTemplateParameters
Value:
ABDataType, ABDataType, AComputeType, AccDataType, CShuffleDataType, DsDataType, EDataType, \
BElementwiseOp, AElementwiseOp, CDEElementwiseOp, NumGemmKPrefetchStage, BlockSize, \
NPerBlock, MPerBlock, KPerBlock, BK1, AK1, NPerXDL, MPerXDL, NXdlPerWave_, MXdlPerWave, \
BBlockTransferThreadClusterLengths_BK0_N_BK1, BBlockTransferThreadClusterArrangeOrder, \
BBlockTransferSrcAccessOrder, BBlockTransferSrcVectorDim, \
BBlockTransferSrcScalarPerVector, BBlockTransferDstScalarPerVector_BK1, false, \
BBlockLdsExtraN, ABlockTransferThreadClusterLengths_AK0_M_AK1, \
ABlockTransferThreadClusterArrangeOrder, ABlockTransferSrcAccessOrder, \
ABlockTransferSrcVectorDim, ABlockTransferSrcScalarPerVector, \
ABlockTransferDstScalarPerVector_AK1, false, ABlockLdsExtraM, \
CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, \
CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, \
CDEBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVersion::v1, BComputeType

◆ GridwiseGemmMultiDTemplateParams

#define GridwiseGemmMultiDTemplateParams
Value:
ABDataType, ABDataType, AComputeType, AccDataType, CShuffleDataType, DsDataType, EDataType, \
AElementwiseOp, BElementwiseOp, CDEElementwiseOp, NumGemmKPrefetchStage, BlockSize, \
MPerBlock, NPerBlock, KPerBlock, AK1, BK1, MPerXDL, NPerXDL, MXdlPerWave, NXdlPerWave_, \
ABlockTransferThreadClusterLengths_AK0_M_AK1, ABlockTransferThreadClusterArrangeOrder, \
ABlockTransferSrcAccessOrder, ABlockTransferSrcVectorDim, \
ABlockTransferSrcScalarPerVector, ABlockTransferDstScalarPerVector_AK1, false, \
ABlockLdsExtraM, BBlockTransferThreadClusterLengths_BK0_N_BK1, \
BBlockTransferThreadClusterArrangeOrder, BBlockTransferSrcAccessOrder, \
BBlockTransferSrcVectorDim, BBlockTransferSrcScalarPerVector, \
BBlockTransferDstScalarPerVector_BK1, false, BBlockLdsExtraN, \
CShuffleMXdlPerWavePerShuffle, CShuffleNXdlPerWavePerShuffle, \
CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock, \
CDEBlockTransferScalarPerVector_NPerBlock, LoopSched, PipelineVersion::v1, BComputeType