#include <iostream>

#include <hipdnn_data_sdk/utilities/Tensor.hpp>
#include <hipdnn_data_sdk/utilities/Workspace.hpp>
#include <hipdnn_frontend.hpp>

#include "utils.hpp"

int main()
{
    using InputType = float;

    const int64_t n = 4; // Batch size

    // Input
    const int64_t c = 32; // Number of channels
    const int64_t h = 16; // Height
    const int64_t w = 16; // Width

    // Filter
    const int64_t k = 64; // Number of filters
    const int64_t r = 3; // Height
    const int64_t s = 3; // Width

    // Conv param
    const int64_t strideH = 1; // Height stride
    const int64_t strideW = 1; // Width stride
    const int64_t padH = 1; // Height padding
    const int64_t padW = 1; // Width padding
    const int64_t dilH = 1; // Height dilation
    const int64_t dilW = 1; // Width dilation

    auto buildConvBackwardGraph = [=](hipdnnHandle_t handle) {
        auto graph = std::make_shared<hipdnn_frontend::graph::Graph>();
        graph->set_name("conv_backward_graph")
            .set_io_data_type(hipdnn_frontend::getDataTypeEnumFromType<InputType>())
            .set_intermediate_data_type(hipdnn_frontend::getDataTypeEnumFromType<InputType>())
            .set_compute_data_type(hipdnn_frontend::DataType::FLOAT);

        auto loss = std::make_shared<hipdnn_frontend::graph::TensorAttributes>(
            hipdnn_frontend::graph::Tensor_attributes()
                .set_name("loss")
                .set_dim({n, k, h, w})
                .set_stride({k * h * w, 1, k * w, k}));

        auto filter = std::make_shared<hipdnn_frontend::graph::TensorAttributes>(
            hipdnn_frontend::graph::Tensor_attributes()
                .set_name("filter")
                .set_dim({k, c, r, s})
                .set_stride({c * r * s, 1, c * s, c}));

        auto convDgradAttributes = hipdnn_frontend::graph::ConvDgradAttributes()
                                       .set_name("conv_backward_node")
                                       .set_padding({padH, padW})
                                       .set_stride({strideH, strideW})
                                       .set_dilation({dilH, dilW});
        auto dx = graph->conv_dgrad(loss, filter, convDgradAttributes);
        dx->set_output(true);

        // build graph
        HIPDNN_FE_CHECK(graph->build(handle));

        return std::make_tuple(graph, loss, filter, dx);
    };

    auto backend = hipdnn_frontend::detail::hipdnnBackend();
    if(!backend)
    {
        std::cout << "Creat backend failed. \n";
        return 1;
    }

    hipdnnHandle_t handle;
    HIPDNN_CHECK(backend->create(&handle));

    auto [graph, loss, filter, dx] = buildConvBackwardGraph(handle);

    hipdnn_data_sdk::utilities::Tensor<InputType> lossTensor(loss->get_dim(), loss->get_stride());
    hipdnn_data_sdk::utilities::Tensor<InputType> filterTensor(filter->get_dim(),
                                                               filter->get_stride());
    hipdnn_data_sdk::utilities::Tensor<InputType> dxTensor(dx->get_dim(), dx->get_stride());

    std::unordered_map<int64_t, void*> variantPack;
    variantPack[loss->get_uid()] = lossTensor.memory().deviceData();
    variantPack[filter->get_uid()] = filterTensor.memory().deviceData();
    variantPack[dx->get_uid()] = dxTensor.memory().deviceData();

    int64_t workspaceSize = 0;
    HIPDNN_FE_CHECK(graph->get_workspace_size(workspaceSize));
    const hipdnn_data_sdk::utilities::Workspace workspace(static_cast<size_t>(workspaceSize));

    HIPDNN_FE_CHECK(graph->execute(handle, variantPack, workspace.get()));

    std::cout << "Convolution backward graph execution complete. \n";

    HIPDNN_CHECK(backend->destroy(handle));
    return 0;
}