ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Example for conv2d backward weight fp16 (#106)

Browse Source 
* add wrw reference

* start device

* raw not split version

* run simple example

* start to use atomic add

* simple transform result correct

* first version that can run

* fix atomic and set operator choice

* add check split-k

* format

* change input parameter

* add pad for t total

* rename example index

Co-authored-by: ltqin <letaoqin@amd.com>
This commit is contained in:
ltqin
2022-03-05 11:18:15 +08:00
committed by GitHub
parent 7e9a9d32c7
commit 7a9b93f4b6
6 changed files with 1284 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

710
device_operation/include/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp Normal file
View File

@@ -0,0 +1,710 @@
#ifndef DEVICE_CONV2D_WRW_XDL_C_SHUFFLE_NHWC_KYXC_NHWK_HPP
#define DEVICE_CONV2D_WRW_XDL_C_SHUFFLE_NHWC_KYXC_NHWK_HPP
#include <iostream>
#include <sstream>
#include "device.hpp"
#include "device_base.hpp"
#include "device_conv_backward_weight.hpp"
#include "convolution_forward_specialization.hpp"
#include "common_header.hpp"
#include "tensor_layout.hpp"
#include "tensor_descriptor.hpp"
#include "tensor_descriptor_helper.hpp"
#include "gridwise_gemm_xdlops_v2r4r2.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
// out[N, Ho, Wo, K] = in[N, Hi, Wi, C] * wei[K, Y, X, C]
template <typename InDataType,
typename WeiDataType,
typename OutDataType,
typename AccDataType,
typename InElementwiseOperation,
typename WeiElementwiseOperation,
typename OutElementwiseOperation,
ck::index_t BlockSize,
ck::index_t MPerBlock,
ck::index_t NPerBlock,
ck::index_t K0PerBlock,
ck::index_t K1,
ck::index_t MPerXdl,
ck::index_t NPerXdl,
ck::index_t MXdlPerWave,
ck::index_t NXdlPerWave,
typename ABlockTransferThreadClusterLengths_K0_M_K1,
typename ABlockTransferThreadClusterArrangeOrder,
typename ABlockTransferSrcAccessOrder,
ck::index_t ABlockTransferSrcVectorDim,
ck::index_t ABlockTransferSrcScalarPerVector,
ck::index_t ABlockTransferDstScalarPerVector_K1,
bool ABlockLdsAddExtraM,
typename BBlockTransferThreadClusterLengths_K0_N_K1,
typename BBlockTransferThreadClusterArrangeOrder,
typename BBlockTransferSrcAccessOrder,
ck::index_t BBlockTransferSrcVectorDim,
ck::index_t BBlockTransferSrcScalarPerVector,
ck::index_t BBlockTransferDstScalarPerVector_K1,
bool BBlockLdsAddExtraN,
index_t CShuffleMXdlPerWavePerShuffle,
index_t CShuffleNXdlPerWavePerShuffle,
typename CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
index_t CBlockTransferScalarPerVector_NWaveNPerXdl>
struct DeviceConv2dWrWXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
: public DeviceConvWrw<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>
{
using DeviceOp = DeviceConv2dWrWXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K;
using ADataType = OutDataType;
using BDataType = InDataType;
using CDataType = WeiDataType;
using AElementwiseOperation = OutElementwiseOperation;
using BElementwiseOperation = InElementwiseOperation;
using CElementwiseOperation = WeiElementwiseOperation;
// TODO make A/B datatype different
using ABDataType = InDataType;
static constexpr index_t NDimSpatial = 2;
static constexpr auto I0 = Number<0>{};
static constexpr auto I1 = Number<1>{};
static constexpr auto I2 = Number<2>{};
static constexpr auto I3 = Number<3>{};
static constexpr auto I4 = Number<4>{};
static constexpr auto I5 = Number<5>{};
static constexpr auto K1Number = Number<K1>{};
static constexpr auto GemmK1Number = K1Number;
static auto
MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(ck::index_t N,
ck::index_t K,
ck::index_t C,
std::vector<ck::index_t> input_spatial_lengths,
std::vector<ck::index_t> filter_spatial_lengths,
std::vector<ck::index_t> output_spatial_lengths,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
ck::index_t batch_k)
{
using namespace ck;
const index_t Hi = input_spatial_lengths[0];
const index_t Wi = input_spatial_lengths[1];
const index_t Ho = output_spatial_lengths[0];
const index_t Wo = output_spatial_lengths[1];
const index_t Y = filter_spatial_lengths[0];
const index_t X = filter_spatial_lengths[1];
const index_t ConvStrideH = conv_filter_strides[0];
const index_t ConvStrideW = conv_filter_strides[1];
const index_t ConvDilationH = conv_filter_dilations[0];
const index_t ConvDilationW = conv_filter_dilations[1];
const index_t InLeftPadH = input_left_pads[0];
const index_t InLeftPadW = input_left_pads[1];
const index_t InRightPadH = input_right_pads[0];
const index_t InRightPadW = input_right_pads[1];
const index_t GemmKTotal = N * Ho * Wo;
const index_t GemmM = K;
const index_t GemmN = C * X * Y;
const index_t GemmKBatch = batch_k;
const index_t GemmK0 =
math::integer_divide_ceil(GemmKTotal, GemmK1Number * K0PerBlock * GemmKBatch) *
K0PerBlock;
const index_t GemmKPad = GemmKBatch * GemmK0 * GemmK1Number;
const auto out_gemmktotal_gemmm_grid_desc =
make_naive_tensor_descriptor_packed(make_tuple(N * Ho * Wo, K));
const auto in_n_hi_wi_c_grid_desc =
make_naive_tensor_descriptor_packed(make_tuple(N, Hi, Wi, C));
// A: output tensor
const auto out_gemmkpad_gemmm_grid_desc = transform_tensor_descriptor(
out_gemmktotal_gemmm_grid_desc,
make_tuple(make_right_pad_transform(GemmKTotal, GemmKPad - GemmKTotal),
make_pass_through_transform(GemmM)),
make_tuple(Sequence<0>{}, Sequence<1>{}),
make_tuple(Sequence<0>{}, Sequence<1>{}));
const auto out_gemmkbatch_gemmk0_gemmm_gemmk1_grid_desc = transform_tensor_descriptor(
out_gemmkpad_gemmm_grid_desc,
make_tuple(make_unmerge_transform(make_tuple(GemmKBatch, GemmK0, GemmK1Number)),
make_pass_through_transform(GemmM)),
make_tuple(Sequence<0>{}, Sequence<1>{}),
make_tuple(Sequence<0, 1, 3>{}, Sequence<2>{}));
// B: input tensor
const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
in_n_hi_wi_c_grid_desc,
make_tuple(make_pass_through_transform(N),
make_pad_transform(Hi, InLeftPadH, InRightPadH),
make_pad_transform(Wi, InLeftPadW, InRightPadW),
make_pass_through_transform(C)),
make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
const auto in_n_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
in_n_hip_wip_c_grid_desc,
make_tuple(
make_pass_through_transform(N),
make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
make_pass_through_transform(C)),
make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3, 4>{}, Sequence<5>{}));
const auto in_gemmktotal_gemmn_grid_desc =
transform_tensor_descriptor(in_n_y_ho_x_wo_c_grid_desc,
make_tuple(make_merge_transform(make_tuple(Y, X, C)),
make_merge_transform(make_tuple(N, Ho, Wo))),
make_tuple(Sequence<1, 3, 5>{}, Sequence<0, 2, 4>{}),
make_tuple(Sequence<1>{}, Sequence<0>{}));
const auto in_gemmkpad_gemmn_grid_desc = transform_tensor_descriptor(
in_gemmktotal_gemmn_grid_desc,
make_tuple(make_right_pad_transform(GemmKTotal, GemmKPad - GemmKTotal),
make_pass_through_transform(GemmN)),
make_tuple(Sequence<0>{}, Sequence<1>{}),
make_tuple(Sequence<0>{}, Sequence<1>{}));
const auto in_gemmkbatch_gemmk0_gemmn_gemmk1_grid_desc = transform_tensor_descriptor(
in_gemmkpad_gemmn_grid_desc,
make_tuple(make_unmerge_transform(make_tuple(GemmKBatch, GemmK0, GemmK1Number)),
make_pass_through_transform(GemmN)),
make_tuple(Sequence<0>{}, Sequence<1>{}),
make_tuple(Sequence<0, 1, 3>{}, Sequence<2>{}));
// C: weight tensor
const auto wei_gemmm_gemmn_grid_desc =
make_naive_tensor_descriptor_packed(make_tuple(K, Y * X * C));
return make_tuple(out_gemmkbatch_gemmk0_gemmm_gemmk1_grid_desc,
in_gemmkbatch_gemmk0_gemmn_gemmk1_grid_desc,
wei_gemmm_gemmn_grid_desc);
}
using ABCGridDescs = decltype(MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(
1, 1, 1, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, 1));
using AGridDesc_K0_M_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I0])>;
using BGridDesc_K0_N_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I1])>;
using CGridDesc_M_N = remove_cvref_t<decltype(ABCGridDescs{}[I2])>;
// GridwiseGemm
using GridwiseGemm = GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_v2r4r2<
BlockSize,
ADataType, // TODO: distinguish A/B datatype
AccDataType,
CDataType,
InMemoryDataOperationEnum_t::Set,
AGridDesc_K0_M_K1,
BGridDesc_K0_N_K1,
CGridDesc_M_N,
AElementwiseOperation,
BElementwiseOperation,
CElementwiseOperation,
MPerBlock,
NPerBlock,
K0PerBlock,
MPerXdl,
NPerXdl,
K1,
MXdlPerWave,
NXdlPerWave,
ABlockTransferThreadClusterLengths_K0_M_K1,
ABlockTransferThreadClusterArrangeOrder,
ABlockTransferSrcAccessOrder,
ABlockTransferSrcVectorDim,
ABlockTransferSrcScalarPerVector,
ABlockTransferDstScalarPerVector_K1,
false, // AThreadTransferSrcResetCoordinateAfterRun,
ABlockLdsAddExtraM,
BBlockTransferThreadClusterLengths_K0_N_K1,
BBlockTransferThreadClusterArrangeOrder,
BBlockTransferSrcAccessOrder,
BBlockTransferSrcVectorDim,
BBlockTransferSrcScalarPerVector,
BBlockTransferDstScalarPerVector_K1,
false, // BThreadTransferSrcResetCoordinateAfterRun,
BBlockLdsAddExtraN,
CShuffleMXdlPerWavePerShuffle,
CShuffleNXdlPerWavePerShuffle,
CBlockTransferScalarPerVector_NWaveNPerXdl,
CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock>;
using GridwiseGemmAtomicAdd = GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_v2r4r2<
BlockSize,
ADataType, // TODO: distinguish A/B datatype
AccDataType,
CDataType,
InMemoryDataOperationEnum_t::AtomicAdd,
AGridDesc_K0_M_K1,
BGridDesc_K0_N_K1,
CGridDesc_M_N,
AElementwiseOperation,
BElementwiseOperation,
CElementwiseOperation,
MPerBlock,
NPerBlock,
K0PerBlock,
MPerXdl,
NPerXdl,
K1,
MXdlPerWave,
NXdlPerWave,
ABlockTransferThreadClusterLengths_K0_M_K1,
ABlockTransferThreadClusterArrangeOrder,
ABlockTransferSrcAccessOrder,
ABlockTransferSrcVectorDim,
ABlockTransferSrcScalarPerVector,
ABlockTransferDstScalarPerVector_K1,
false, // AThreadTransferSrcResetCoordinateAfterRun,
ABlockLdsAddExtraM,
BBlockTransferThreadClusterLengths_K0_N_K1,
BBlockTransferThreadClusterArrangeOrder,
BBlockTransferSrcAccessOrder,
BBlockTransferSrcVectorDim,
BBlockTransferSrcScalarPerVector,
BBlockTransferDstScalarPerVector_K1,
false, // BThreadTransferSrcResetCoordinateAfterRun,
BBlockLdsAddExtraN,
CShuffleMXdlPerWavePerShuffle,
CShuffleNXdlPerWavePerShuffle,
CBlockTransferScalarPerVector_NWaveNPerXdl,
CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock>;
// Argument
using CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock =
decltype(GridwiseGemm::MakeCGridDesc_MBlock_MPerBlock_NBlock_NPerBlock(CGridDesc_M_N{}));
using Block2CTileMap =
decltype(GridwiseGemm::MakeCBlockClusterAdaptor(CGridDesc_M_N{}, 1, 1, 1));
struct Argument : public BaseArgument
{
Argument(const InDataType* p_in_grid,
WeiDataType* p_wei_grid,
const OutDataType* p_out_grid,
ck::index_t N,
ck::index_t K,
ck::index_t C,
std::vector<ck::index_t> input_spatial_lengths,
std::vector<ck::index_t> filter_spatial_lengths,
std::vector<ck::index_t> output_spatial_lengths,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
ck::index_t M01,
ck::index_t N01,
InElementwiseOperation in_element_op,
WeiElementwiseOperation wei_element_op,
OutElementwiseOperation out_element_op,
ck::index_t split_k)
: p_a_grid_{p_out_grid},
p_b_grid_{p_in_grid},
p_c_grid_{p_wei_grid},
a_grid_desc_kbatch_k0_m_k1_{},
b_grid_desc_kbatch_k0_n_k1_{},
c_grid_desc_m_n_{},
c_grid_desc_mblock_mperblock_nblock_nperblock_{},
block_2_ctile_map_{},
M01_{M01},
N01_{N01},
a_element_op_{out_element_op},
b_element_op_{in_element_op},
c_element_op_{wei_element_op},
Conv_N_{N},
Conv_K_{K},
Conv_C_{C},
output_spatial_lengths_{output_spatial_lengths},
filter_spatial_lengths_{filter_spatial_lengths},
conv_filter_strides_{conv_filter_strides},
input_left_pads_{input_left_pads},
input_right_pads_{input_right_pads},
k_batch_{split_k}
{
const auto descs =
DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(N,
K,
C,
input_spatial_lengths,
filter_spatial_lengths,
output_spatial_lengths,
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
k_batch_);
a_grid_desc_kbatch_k0_m_k1_ = descs[I0];
b_grid_desc_kbatch_k0_n_k1_ = descs[I1];
c_grid_desc_m_n_ = descs[I2];
if(GridwiseGemm::CheckValidity(a_grid_desc_kbatch_k0_m_k1_,
b_grid_desc_kbatch_k0_n_k1_,
c_grid_desc_m_n_,
M01_,
N01_))
{
c_grid_desc_mblock_mperblock_nblock_nperblock_ =
GridwiseGemm::MakeCGridDesc_MBlock_MPerBlock_NBlock_NPerBlock(c_grid_desc_m_n_);
block_2_ctile_map_ =
GridwiseGemm::MakeCBlockClusterAdaptor(c_grid_desc_m_n_, M01, N01, k_batch_);
}
}
const ADataType* p_a_grid_;
const BDataType* p_b_grid_;
CDataType* p_c_grid_;
AGridDesc_K0_M_K1 a_grid_desc_kbatch_k0_m_k1_;
BGridDesc_K0_N_K1 b_grid_desc_kbatch_k0_n_k1_;
CGridDesc_M_N c_grid_desc_m_n_;
CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock c_grid_desc_mblock_mperblock_nblock_nperblock_;
Block2CTileMap block_2_ctile_map_;
index_t M01_;
index_t N01_;
InElementwiseOperation a_element_op_;
OutElementwiseOperation b_element_op_;
WeiElementwiseOperation c_element_op_;
// for checking IsSupportedArgument()
index_t Conv_N_;
index_t Conv_K_;
index_t Conv_C_;
std::vector<index_t> output_spatial_lengths_;
std::vector<index_t> filter_spatial_lengths_;
std::vector<index_t> conv_filter_strides_;
std::vector<index_t> input_left_pads_;
std::vector<index_t> input_right_pads_;
index_t k_batch_;
};
// Invoker
struct Invoker : public BaseInvoker
{
using Argument = DeviceOp::Argument;
void ShowInfo(const Argument& arg)
{
std::cout << "arg.a_grid_desc_kbatch_k0_m_k1_{"
<< arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I0) << ", "
<< arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I1) << ", "
<< arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I2) << ", "
<< arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I3) << "}" << std::endl;
std::cout << "arg.b_grid_desc_kbatch_k0_n_k1_{"
<< arg.b_grid_desc_kbatch_k0_n_k1_.GetLength(I0) << ", "
<< arg.b_grid_desc_kbatch_k0_n_k1_.GetLength(I1) << ", "
<< arg.b_grid_desc_kbatch_k0_n_k1_.GetLength(I2) << ", "
<< arg.b_grid_desc_kbatch_k0_n_k1_.GetLength(I3) << "}" << std::endl;
std::cout << "arg.c_grid_desc_m_n_{ " << arg.c_grid_desc_m_n_.GetLength(I0) << ", "
<< arg.c_grid_desc_m_n_.GetLength(I1) << "}" << std::endl;
}
float Run(const Argument& arg, int nrepeat = 1)
{
ShowInfo(arg);
if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_kbatch_k0_m_k1_,
arg.b_grid_desc_kbatch_k0_n_k1_,
arg.c_grid_desc_m_n_,
arg.M01_,
arg.N01_))
{
throw std::runtime_error(
"wrong! GridwiseGemm_km_kn_m0m1n0n1_xdlops_v3r1 has invalid setting");
}
const auto kbatch = arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I0);
const index_t grid_size = GridwiseGemm::CalculateGridSize(arg.c_grid_desc_m_n_, kbatch);
const auto K0 = arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I1);
const bool has_main_k0_block_loop = GridwiseGemm::CalculateHasMainK0BlockLoop(K0);
float ave_time = 0;
const auto Run = [&](const auto& kernel) {
if(nrepeat > 0)
{
ave_time =
launch_and_time_kernel(kernel,
nrepeat,
dim3(grid_size),
dim3(BlockSize),
0,
arg.p_a_grid_,
arg.p_b_grid_,
arg.p_c_grid_,
arg.a_grid_desc_kbatch_k0_m_k1_,
arg.b_grid_desc_kbatch_k0_n_k1_,
arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
arg.a_element_op_,
arg.b_element_op_,
arg.c_element_op_,
arg.block_2_ctile_map_);
}
if(kbatch > 1 || nrepeat <= 0)
{
hipGetErrorString(hipMemset(
arg.p_c_grid_,
0,
arg.c_grid_desc_mblock_mperblock_nblock_nperblock_.GetElementSpaceSize() *
sizeof(CDataType)));
launch_kernel(kernel,
dim3(grid_size),
dim3(BlockSize),
0,
arg.p_a_grid_,
arg.p_b_grid_,
arg.p_c_grid_,
arg.a_grid_desc_kbatch_k0_m_k1_,
arg.b_grid_desc_kbatch_k0_n_k1_,
arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
arg.a_element_op_,
arg.b_element_op_,
arg.c_element_op_,
arg.block_2_ctile_map_);
}
};
if(has_main_k0_block_loop)
{
if(kbatch == 1)
{
const auto kernel = kernel_gemm_xdlops_v2r4r2<
GridwiseGemm,
ADataType, // TODO: distiguish A/B datatype
CDataType,
remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
remove_reference_t<DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
OutElementwiseOperation,
InElementwiseOperation,
WeiElementwiseOperation,
remove_reference_t<DeviceOp::Block2CTileMap>,
true>;
Run(kernel);
}
else
{
const auto kernel = kernel_gemm_xdlops_v2r4r2<
GridwiseGemmAtomicAdd,
ADataType, // TODO: distiguish A/B datatype
CDataType,
remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
remove_reference_t<DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
OutElementwiseOperation,
InElementwiseOperation,
WeiElementwiseOperation,
remove_reference_t<DeviceOp::Block2CTileMap>,
true>;
Run(kernel);
}
}
else
{
if(kbatch == 1)
{
const auto kernel = kernel_gemm_xdlops_v2r4r2<
GridwiseGemm,
ADataType, // TODO: distiguish A/B datatype
CDataType,
remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
remove_reference_t<DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
OutElementwiseOperation,
InElementwiseOperation,
WeiElementwiseOperation,
remove_reference_t<DeviceOp::Block2CTileMap>,
false>;
Run(kernel);
}
else
{
const auto kernel = kernel_gemm_xdlops_v2r4r2<
GridwiseGemmAtomicAdd,
ADataType, // TODO: distiguish A/B datatype
CDataType,
remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
remove_reference_t<DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
OutElementwiseOperation,
InElementwiseOperation,
WeiElementwiseOperation,
remove_reference_t<DeviceOp::Block2CTileMap>,
false>;
Run(kernel);
}
}
return ave_time;
}
float Run(const BaseArgument* p_arg, int nrepeat = 1) override
{
return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat);
}
};
static constexpr bool IsValidCompilationParameter()
{
// TODO: properly implement this check
return true;
}
static bool IsSupportedArgument(const Argument& arg)
{
// vector load A/B matrix from global memory
if(!(ABlockTransferSrcVectorDim == 2 && BBlockTransferSrcVectorDim == 2 &&
arg.Conv_K_ % ABlockTransferSrcScalarPerVector == 0 &&
arg.Conv_C_ % BBlockTransferSrcScalarPerVector == 0))
{
return false;
}
// vector store C matrix into global memory
if(!(arg.Conv_C_ % CBlockTransferScalarPerVector_NWaveNPerXdl == 0))
{
return false;
}
// Gridwise GEMM size
return GridwiseGemm::CheckValidity(arg.a_grid_desc_kbatch_k0_m_k1_,
arg.b_grid_desc_kbatch_k0_n_k1_,
arg.c_grid_desc_m_n_,
arg.M01_,
arg.N01_);
}
bool IsSupportedArgument(const BaseArgument* p_arg) override
{
return IsSupportedArgument(*dynamic_cast<const Argument*>(p_arg));
}
static auto MakeArgument(const InDataType* p_in_grid,
WeiDataType* p_wei_grid,
const OutDataType* p_out_grid,
ck::index_t N,
ck::index_t K,
ck::index_t C,
std::vector<ck::index_t> input_spatial_lengths,
std::vector<ck::index_t> filter_spatial_lengths,
std::vector<ck::index_t> output_spatial_lengths,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
InElementwiseOperation in_element_op,
WeiElementwiseOperation wei_element_op,
OutElementwiseOperation out_element_op,
ck::index_t split_k)
{
return Argument{p_in_grid,
p_wei_grid,
p_out_grid,
N,
K,
C,
input_spatial_lengths,
filter_spatial_lengths,
output_spatial_lengths,
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
1,
1,
in_element_op,
wei_element_op,
out_element_op,
split_k};
}
static auto MakeInvoker() { return Invoker{}; }
std::unique_ptr<BaseArgument>
MakeArgumentPointer(const void* p_in_grid,
void* p_wei_grid,
const void* p_out_grid,
ck::index_t N,
ck::index_t K,
ck::index_t C,
std::vector<ck::index_t> input_spatial_lengths,
std::vector<ck::index_t> filter_spatial_lengths,
std::vector<ck::index_t> output_spatial_lengths,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
InElementwiseOperation in_element_op,
WeiElementwiseOperation wei_element_op,
OutElementwiseOperation out_element_op,
ck::index_t split_k) override
{
return std::make_unique<Argument>(static_cast<const InDataType*>(p_in_grid),
static_cast<WeiDataType*>(p_wei_grid),
static_cast<const OutDataType*>(p_out_grid),
N,
K,
C,
input_spatial_lengths,
filter_spatial_lengths,
output_spatial_lengths,
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
1,
1,
in_element_op,
wei_element_op,
out_element_op,
split_k);
}
std::unique_ptr<BaseInvoker> MakeInvokerPointer() override
{
return std::make_unique<Invoker>(Invoker{});
}
std::string GetTypeString() const override
{
auto str = std::stringstream();
// clang-format off
str << "DeviceConv2dWrWXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K"
<< "<"
<< BlockSize << ", "
<< MPerBlock << ", "
<< NPerBlock << ", "
<< K0PerBlock
<< ">";
// clang-format on
return str.str();
}
};
} // namespace device
} // namespace tensor_operation
} // namespace ck
#endif

47
device_operation/include/device_conv_backward_weight.hpp Normal file
View File

@@ -0,0 +1,47 @@
#ifndef DEVICE_CONV_WRW_HPP
#define DEVICE_CONV_WRW_HPP
#include <iostream>
#include "device_base.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
template <typename InElementwiseOperation,
typename WeiElementwiseOperation,
typename OutElementwiseOperation>
struct DeviceConvWrw : public BaseOperator
{
virtual std::unique_ptr<BaseArgument>
MakeArgumentPointer(const void* p_in,
void* p_wei,
const void* p_out,
ck::index_t N,
ck::index_t K,
ck::index_t C,
std::vector<ck::index_t> input_spatial_lengths,
std::vector<ck::index_t> filter_spatial_lengths,
std::vector<ck::index_t> output_spatial_lengths,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
InElementwiseOperation in_element_op,
WeiElementwiseOperation wei_element_op,
OutElementwiseOperation out_element_op,
ck::index_t split_k) = 0;
virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
};
template <typename InElementwiseOperation,
typename WeiElementwiseOperation,
typename OutElementwiseOperation>
using DeviceConvWrwPtr = std::unique_ptr<
DeviceConvWrw<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>>;
} // namespace device
} // namespace tensor_operation
} // namespace ck
#endif

58
example/13_conv2d_backward_weight_xdl/README.md Normal file
View File

@@ -0,0 +1,58 @@
# Instructions for ```conv2d_wrw_xdl``` Example
## Docker script
```bash
docker run \
-it \
--rm \
--privileged \
--group-add sudo \
-w /root/workspace \
-v ${PATH_TO_LOCAL_WORKSPACE}:/root/workspace \
rocm/tensorflow:rocm4.3.1-tf2.6-dev \
/bin/bash
```
## Build ```conv2d_wrw_xdl```
```bash
mkdir build && cd build
```
```bash
# Need to specify target ID, example below is gfx908
cmake \
-D BUILD_DEV=OFF \
-D CMAKE_BUILD_TYPE=Release \
-D CMAKE_CXX_FLAGS="-DCK_AMD_GPU_GFX908 --amdgpu-target=gfx908 -O3 " \
-D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc \
-D CMAKE_PREFIX_PATH=/opt/rocm \
..
```
```bash
make -j conv2d_wrw_xdl
```
## Run ```conv2d_wrw_xdl```
```bash
#arg1: verification (0=no, 1=yes)
#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
#arg3: run kernel # of times (>1)
#arg4: is show log (0=no, 1=yes)
#arg5 to 19: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, RightPx, split-k
./example/conv2d_fwd_xdl 0 1 5 0 4
```
Result
```
in_n_c_hi_wi: dim 4, lengths {128, 1024, 14, 14}, strides {200704, 1, 14336, 1024}
wei_k_c_y_x: dim 4, lengths {256, 1024, 3, 3}, strides {9216, 1, 3072, 1024}
out_n_k_ho_wo: dim 4, lengths {128, 256, 6, 6}, strides {9216, 1, 1536, 256}
arg.a_grid_desc_kbatch_k0_m_k1_{4, 144, 256, 8}
arg.b_grid_desc_kbatch_k0_n_k1_{4, 144, 9216, 8}
arg.c_grid_desc_m_n_{ 256, 9216}
launch_and_time_kernel: grid_dim {576, 1, 1}, block_dim {256, 1, 1}
Warm up
Start running 5 times...
Perf: 0.401084 ms, 54.2112 TFlops, 145.75 GB/s
```

289
example/13_conv2d_backward_weight_xdl/main.cpp Normal file
View File

@@ -0,0 +1,289 @@
#include <iostream>
#include <numeric>
#include <initializer_list>
#include <cstdlib>
#include <stdlib.h>
#include <half.hpp>
#include "config.hpp"
#include "print.hpp"
#include "device.hpp"
#include "host_tensor.hpp"
#include "host_tensor_generator.hpp"
#include "device_tensor.hpp"
#include "tensor_layout.hpp"
#include "element_wise_operation.hpp"
#include "device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
#include "reference_conv_backward_weight.hpp"
using InDataType = ck::half_t;
using WeiDataType = ck::half_t;
using OutDataType = ck::half_t;
using AccDataType = float;
template <ck::index_t... Is>
using S = ck::Sequence<Is...>;
using InLayout = ck::tensor_layout::convolution::NHWC;
using WeiLayout = ck::tensor_layout::convolution::KYXC;
using OutLayout = ck::tensor_layout::convolution::NHWK;
using InElementOp = ck::tensor_operation::element_wise::PassThrough;
using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
// clang-format off
using DeviceConvWrWInstance = ck::tensor_operation::device::
DeviceConv2dWrWXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
InDataType, // InDataType
WeiDataType, // WeiDataType
OutDataType, // OutDataType
AccDataType, // AccDataType
InElementOp, // InElementwiseOperation
WeiElementOp, // WeiElementwiseOperation
OutElementOp, // OutElementwiseOperation
256, // BlockSize
128, // MPerBlock
128, // NPerBlock
4, // K0PerBlock
8, // K1
32, // MPerXdl
32, // NPerXdl
2, // MXdlPerWave
2, // NXdlPerWave
S<1, 4, 16, 4>, // ABlockTransferThreadClusterLengths_K0_M_K1
S<0, 3, 1, 2>, // ABlockTransferThreadClusterArrangeOrder
S<0, 2, 1, 3>, // ABlockTransferSrcAccessOrder
2, // ABlockTransferSrcVectorDim
8, // ABlockTransferSrcScalarPerVector
2, // ABlockTransferDstScalarPerVector_K1
true, // ABlockLdsAddExtraM
S<1, 4, 16, 4>, // BBlockTransferThreadClusterLengths_K0_N_K1
S<0, 3, 1, 2>, // BBlockTransferThreadClusterArrangeOrder
S<0, 2, 1, 3>, // BBlockTransferSrcAccessOrder
2, // BBlockTransferSrcVectorDim
8, // BBlockTransferSrcScalarPerVector
2, // BBlockTransferDstScalarPerVector_K1
true, // BBlockLdsAddExtraN
1, // CShuffleMXdlPerWavePerShuffle
1, // CShuffleNXdlPerWavePerShuffle
S<1, 32, 1, 4>, // CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock
8>; // CBlockTransferScalarPerVector_NWaveNPerXdl
// clang-format on
using ReferenceConvWrwInstance = ck::tensor_operation::host::
ReferenceConvWrw<InDataType, WeiDataType, OutDataType, InElementOp, WeiElementOp, OutElementOp>;
int main(int argc, char* argv[])
{
bool do_verification = 0;
int init_method = 0;
int nrepeat = 5;
int do_log = 0;
int split_k = 4;
// Conv shape
ck::index_t N = 128;
ck::index_t K = 256;
ck::index_t C = 1024;
ck::index_t Y = 3;
ck::index_t X = 3;
ck::index_t Hi = 14;
ck::index_t Wi = 14;
ck::index_t conv_stride_h = 2;
ck::index_t conv_stride_w = 2;
ck::index_t conv_dilation_h = 1;
ck::index_t conv_dilation_w = 1;
ck::index_t in_left_pad_h = 0;
ck::index_t in_left_pad_w = 0;
ck::index_t in_right_pad_h = 0;
ck::index_t in_right_pad_w = 0;
if(argc == 6)
{
do_verification = std::stoi(argv[1]);
init_method = std::stoi(argv[2]);
nrepeat = std::stoi(argv[3]);
do_log = std::stoi(argv[4]);
split_k = std::stoi(argv[5]);
}
else if(argc == 21)
{
do_verification = std::stoi(argv[1]);
init_method = std::stoi(argv[2]);
nrepeat = std::stoi(argv[3]);
do_log = std::stoi(argv[4]);
split_k = std::stoi(argv[5]);
N = std::stoi(argv[6]);
K = std::stoi(argv[7]);
C = std::stoi(argv[8]);
Y = std::stoi(argv[9]);
X = std::stoi(argv[10]);
Hi = std::stoi(argv[11]);
Wi = std::stoi(argv[12]);
conv_stride_h = std::stoi(argv[13]);
conv_stride_w = std::stoi(argv[14]);
conv_dilation_h = std::stoi(argv[15]);
conv_dilation_w = std::stoi(argv[16]);
in_left_pad_h = std::stoi(argv[17]);
in_left_pad_w = std::stoi(argv[18]);
in_right_pad_h = std::stoi(argv[19]);
in_right_pad_w = std::stoi(argv[20]);
}
else
{
printf("arg1: verification (0=no, 1=yes)\n");
printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
printf("arg3: run kernel # of times (>1)\n");
printf("arg4: is show log (0=no, 1=yes)\n");
printf("arg5: split-k \n");
printf("arg6 to 19: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
"RightPx\n");
exit(0);
}
const ck::index_t YEff = (Y - 1) * conv_dilation_h + 1;
const ck::index_t XEff = (X - 1) * conv_dilation_w + 1;
const ck::index_t Ho = (Hi + in_left_pad_h + in_right_pad_h - YEff) / conv_stride_h + 1;
const ck::index_t Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
const std::vector<ck::index_t> conv_filter_strides{{conv_stride_h, conv_stride_w}};
const std::vector<ck::index_t> conv_filter_dilations{{conv_dilation_h, conv_dilation_w}};
const std::vector<ck::index_t> input_left_pads{{in_left_pad_h, in_left_pad_w}};
const std::vector<ck::index_t> input_right_pads{{in_right_pad_h, in_right_pad_w}};
// tensor layout
auto f_host_tensor_descriptor = [](std::size_t N_,
std::size_t C_,
std::size_t H,
std::size_t W,
auto layout) {
if constexpr(ck::is_same<decltype(layout), ck::tensor_layout::convolution::NCHW>::value ||
ck::is_same<decltype(layout), ck::tensor_layout::convolution::KCYX>::value ||
ck::is_same<decltype(layout), ck::tensor_layout::convolution::NKHW>::value)
{
return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
std::vector<std::size_t>({C_ * H * W, H * W, W, 1}));
}
else if constexpr(ck::is_same<decltype(layout),
ck::tensor_layout::convolution::NHWC>::value ||
ck::is_same<decltype(layout),
ck::tensor_layout::convolution::KYXC>::value ||
ck::is_same<decltype(layout),
ck::tensor_layout::convolution::NHWK>::value)
{
return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
std::vector<std::size_t>({C_ * H * W, 1, W * C_, C_}));
}
};
Tensor<InDataType> in_n_c_hi_wi(f_host_tensor_descriptor(N, C, Hi, Wi, InLayout{}));
Tensor<WeiDataType> wei_k_c_y_x_host_result(f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
Tensor<WeiDataType> wei_k_c_y_x_device_result(
f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
Tensor<OutDataType> out_n_k_ho_wo(f_host_tensor_descriptor(N, K, Ho, Wo, OutLayout{}));
std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_host_result.mDesc << std::endl;
std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
switch(init_method)
{
case 0: break;
case 1:
in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
break;
default:
in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_1<InDataType>{1});
out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
}
wei_k_c_y_x_device_result.GenerateTensorValue(GeneratorTensor_1<WeiDataType>{0});
DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
DeviceMem wei_device_buf(sizeof(WeiDataType) *
wei_k_c_y_x_device_result.mDesc.GetElementSpace());
DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
wei_device_buf.ToDevice(wei_k_c_y_x_device_result.mData.data());
// do GEMM
auto conv = DeviceConvWrWInstance{};
auto invoker = conv.MakeInvoker();
auto argument = conv.MakeArgument(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
N,
K,
C,
std::vector<ck::index_t>{{Hi, Wi}},
std::vector<ck::index_t>{{Y, X}},
std::vector<ck::index_t>{{Ho, Wo}},
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
InElementOp{},
WeiElementOp{},
OutElementOp{},
split_k);
if(!conv.IsSupportedArgument(argument))
{
std::cout << "wrong! device_conv with the specified compilation parameters does "
"not support this Conv problem"
<< std::endl;
return 1;
}
float ave_time = invoker.Run(argument, nrepeat);
std::size_t flop = std::size_t(2) * N * K * Ho * Wo * C * Y * X;
std::size_t num_btype = sizeof(InDataType) * (N * C * Hi * Wi) +
sizeof(WeiDataType) * (K * C * Y * X) +
sizeof(OutDataType) * (N * K * Ho * Wo);
float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
float gb_per_sec = num_btype / 1.E6 / ave_time;
std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
<< std::endl;
if(do_verification)
{
auto ref_conv = ReferenceConvWrwInstance{};
auto ref_invoker = ref_conv.MakeInvoker();
auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi,
wei_k_c_y_x_host_result,
out_n_k_ho_wo,
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
InElementOp{},
WeiElementOp{},
OutElementOp{});
ref_invoker.Run(ref_argument);
wei_device_buf.FromDevice(wei_k_c_y_x_device_result.mData.data());
if(do_log)
{
LogRangeAsType<float>(std::cout << "out: ", out_n_k_ho_wo.mData, ",") << std::endl;
LogRangeAsType<float>(std::cout << "in : ", in_n_c_hi_wi.mData, ",") << std::endl;
LogRangeAsType<float>(
std::cout << "wei_device(after): ", wei_k_c_y_x_device_result.mData, ",")
<< std::endl;
LogRangeAsType<float>(std::cout << "wei_host : ", wei_k_c_y_x_host_result.mData, ",")
<< std::endl;
}
check_error(wei_k_c_y_x_host_result, wei_k_c_y_x_device_result);
}
}

3
example/CMakeLists.txt
View File

@@ -24,6 +24,7 @@ set(CONV2D_FWD_XDL_BIAS_RELU_ADD_SOURCE 6_conv2d_fwd_xdl_bias_relu_add/conv2d_fw
set(CONV2D_FWD_XDL_BIAS_RELU_ATOMIC_ADD_SOURCE 7_conv2d_fwd_xdl_bias_relu_atomic_add/conv2d_fwd_xdl_bias_relu_atomic_add.cpp)
set(GEMM_XDL_ALPHA_BETA_SOURCE 8_gemm_xdl_alpha_beta/gemm_xdl_alpha_beta.cpp)
set(CONV2D_FWD_XDL_INT8_SOURCE 9_conv2d_fwd_xdl_int8/conv2d_fwd_xdl_int8.cpp)
set(CONV2D_WRW_XDL_SOURCE 13_conv2d_backward_weight_xdl/main.cpp)
set(CONV3D_FWD_XDL_SOURCE 10_conv3d_fwd_xdl/conv3d_fwd_xdl.cpp)
set(CONVND_FWD_XDL_SOURCE 11_convnd_fwd_xdl/convnd_fwd_xdl.cpp)
set(CONV2D_BWD_DATA_XDL_SOURCE 12_conv2d_bwd_data_xdl/conv2d_bwd_data_xdl.cpp)
@@ -39,6 +40,7 @@ add_executable(conv2d_fwd_xdl_bias_relu_add ${CONV2D_FWD_XDL_BIAS_RELU_ADD_SOURC
add_executable(conv2d_fwd_xdl_bias_relu_atomic_add ${CONV2D_FWD_XDL_BIAS_RELU_ATOMIC_ADD_SOURCE})
add_executable(gemm_xdl_alpha_beta ${GEMM_XDL_ALPHA_BETA_SOURCE})
add_executable(conv2d_fwd_xdl_int8 ${CONV2D_FWD_XDL_INT8_SOURCE})
add_executable(conv2d_wrw_xdl ${CONV2D_WRW_XDL_SOURCE})
add_executable(conv3d_fwd_xdl ${CONV3D_FWD_XDL_SOURCE})
add_executable(convnd_fwd_xdl ${CONVND_FWD_XDL_SOURCE})
add_executable(conv2d_bwd_data_xdl ${CONV2D_BWD_DATA_XDL_SOURCE})
@@ -54,6 +56,7 @@ target_link_libraries(conv2d_fwd_xdl_bias_relu_add PRIVATE host_tensor)
target_link_libraries(conv2d_fwd_xdl_bias_relu_atomic_add PRIVATE host_tensor)
target_link_libraries(gemm_xdl_alpha_beta PRIVATE host_tensor)
target_link_libraries(conv2d_fwd_xdl_int8 PRIVATE host_tensor)
target_link_libraries(conv2d_wrw_xdl PRIVATE host_tensor)
target_link_libraries(conv3d_fwd_xdl PRIVATE host_tensor)
target_link_libraries(convnd_fwd_xdl PRIVATE host_tensor)
target_link_libraries(conv2d_bwd_data_xdl PRIVATE host_tensor)

177
reference_operation/include/reference_conv_backward_weight.hpp Normal file
View File

@@ -0,0 +1,177 @@
#ifndef REFERENCE_CONV_WRW_HPP
#define REFERENCE_CONV_WRW_HPP
#include <iostream>
#include <sstream>
#include "device_base.hpp"
#include "host_tensor.hpp"
namespace ck {
namespace tensor_operation {
namespace host {
// out[N, K, Ho, Wo] = in[N, C, Hi, Wi] * wei[K, C, Y, X]
template <typename InDataType,
typename WeiDataType,
typename OutDataType,
typename InElementwiseOperation,
typename WeiElementwiseOperation,
typename OutElementwiseOperation>
struct ReferenceConvWrw : public device::BaseOperator
{
// Argument
struct Argument : public device::BaseArgument
{
Argument(const Tensor<InDataType>& in_n_c_hi_wi,
Tensor<WeiDataType>& wei_k_c_y_x,
const Tensor<OutDataType>& out_n_k_ho_wo,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
InElementwiseOperation in_element_op,
WeiElementwiseOperation wei_element_op,
OutElementwiseOperation out_element_op)
: in_n_c_hi_wi_{in_n_c_hi_wi},
wei_k_c_y_x_{wei_k_c_y_x},
out_n_k_ho_wo_{out_n_k_ho_wo},
conv_strides_{conv_filter_strides},
conv_dilations_{conv_filter_dilations},
in_left_pads_{input_left_pads},
in_right_pads_{input_right_pads},
in_element_op_{in_element_op},
wei_element_op_{wei_element_op},
out_element_op_{out_element_op}
{
}
const Tensor<InDataType>& in_n_c_hi_wi_;
Tensor<WeiDataType>& wei_k_c_y_x_;
const Tensor<OutDataType>& out_n_k_ho_wo_;
std::vector<index_t> conv_strides_;
std::vector<index_t> conv_dilations_;
std::vector<index_t> in_left_pads_;
std::vector<index_t> in_right_pads_;
InElementwiseOperation in_element_op_;
WeiElementwiseOperation wei_element_op_;
OutElementwiseOperation out_element_op_;
};
// Invoker
struct Invoker : public device::BaseInvoker
{
using Argument = ReferenceConvWrw::Argument;
float Run(const Argument& arg)
{
constexpr auto I0 = Number<0>{};
constexpr auto I1 = Number<1>{};
auto f_kcyx = [&](auto k, auto c, auto y, auto x) {
float v_acc = 0;
for(int n = 0; n < arg.out_n_k_ho_wo_.mDesc.GetLengths()[0]; ++n)
{
for(int ho = 0; ho < arg.out_n_k_ho_wo_.mDesc.GetLengths()[2]; ++ho)
{
int hi = ho * arg.conv_strides_[I0] + y * arg.conv_dilations_[I0] -
arg.in_left_pads_[I0];
for(int wo = 0; wo < arg.out_n_k_ho_wo_.mDesc.GetLengths()[3]; ++wo)
{
int wi = wo * arg.conv_strides_[I1] + x * arg.conv_dilations_[I1] -
arg.in_left_pads_[I1];
if(hi >= 0 && hi < arg.in_n_c_hi_wi_.mDesc.GetLengths()[2] && wi >= 0 &&
wi < arg.in_n_c_hi_wi_.mDesc.GetLengths()[3])
{
float v_out;
float v_in;
arg.out_element_op_(
v_out,
ck::type_convert<float>(arg.out_n_k_ho_wo_(n, k, ho, wo)));
arg.in_element_op_(
v_in, ck::type_convert<float>(arg.in_n_c_hi_wi_(n, c, hi, wi)));
v_acc += v_out * v_in;
}
}
}
}
float v_wei;
arg.wei_element_op_(v_wei, v_acc);
arg.wei_k_c_y_x_(k, c, y, x) = ck::type_convert<OutDataType>(v_wei);
};
make_ParallelTensorFunctor(f_kcyx,
arg.wei_k_c_y_x_.mDesc.GetLengths()[0],
arg.wei_k_c_y_x_.mDesc.GetLengths()[1],
arg.wei_k_c_y_x_.mDesc.GetLengths()[2],
arg.wei_k_c_y_x_.mDesc.GetLengths()[3])(
std::thread::hardware_concurrency());
return 0;
}
float Run(const device::BaseArgument* p_arg, int) override
{
return Run(*dynamic_cast<const Argument*>(p_arg));
}
};
static constexpr bool IsValidCompilationParameter()
{
// TODO: properly implement this check
return true;
}
bool IsSupportedArgument(const device::BaseArgument*) override { return true; }
static auto MakeArgument(const Tensor<InDataType>& in_n_c_hi_wi,
Tensor<WeiDataType>& wei_k_c_y_x,
const Tensor<OutDataType>& out_n_k_ho_wo,
std::vector<ck::index_t> conv_filter_strides,
std::vector<ck::index_t> conv_filter_dilations,
std::vector<ck::index_t> input_left_pads,
std::vector<ck::index_t> input_right_pads,
InElementwiseOperation in_element_op,
WeiElementwiseOperation wei_element_op,
OutElementwiseOperation out_element_op)
{
return Argument{in_n_c_hi_wi,
wei_k_c_y_x,
out_n_k_ho_wo,
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
in_element_op,
wei_element_op,
out_element_op};
}
static auto MakeInvoker() { return Invoker{}; }
virtual std::unique_ptr<device::BaseInvoker> MakeInvokerPointer()
{
return std::make_unique<Invoker>(Invoker{});
}
std::string GetTypeString() const override
{
auto str = std::stringstream();
// clang-format off
str << "ReferenceConvFwd"
<< std::endl;
// clang-format on
return str.str();
}
};
} // namespace host
} // namespace tensor_operation
} // namespace ck
#endif