import hipdnn
import torch


def build_conv_wrw_graph(
    hipdnn_handle, torch_tensor_x, torch_tensor_dy, padding, stride, dilation, hipdnn_data_type
):
    # Create graph
    graph = hipdnn.pygraph(
        handle=hipdnn_handle,
        io_data_type=hipdnn_data_type,
        intermediate_data_type=hipdnn.data_type.FLOAT,
        compute_data_type=hipdnn.data_type.FLOAT,
        name="convolution_wrw",
    )

    # Create hipdnn tensors
    hipdnn_tensor_x = graph.tensor_like(torch_tensor_x)
    hipdnn_tensor_dy = graph.tensor_like(torch_tensor_dy)

    # Create conv op
    hipdnn_tensor_dw = graph.conv_wgrad(
        image=hipdnn_tensor_x,
        loss=hipdnn_tensor_dy,
        padding=padding,
        stride=stride,
        dilation=dilation,
        name="conv2d_wrw",
    )
    hipdnn_tensor_dw.set_output(True)
    graph.build(hipdnn_handle)

    return (graph, hipdnn_tensor_x, hipdnn_tensor_dy, hipdnn_tensor_dw)


if __name__ == "__main__":
    # Input dimensions
    n = 4  # Batch size
    c = 32  # Number of input channels
    h = 16  # Height
    w = 16  # Width

    # Filter dimensions
    k = 64  # Number of output channels
    r = 3  # Filter height
    s = 3  # Filter width

    # Convolution parameters
    stride_h = 1  # Height stride
    stride_w = 1  # Width stride
    pad_h = 1  # Height padding
    pad_w = 1  # Width padding
    dil_h = 1  # Height dilation
    dil_w = 1  # Width dilation

    hipdnn_data_type = hipdnn.data_type.FLOAT
    torch_data_type = torch.float32

    torch_tensor_x = torch.rand(n, c, h, w, dtype=torch_data_type, device="cuda")
    torch_tensor_dy = torch.rand(n, k, h, w, dtype=torch_data_type, device="cuda")

    hipdnn_handle = hipdnn.create_handle()

    graph, hipdnn_tensor_x, hipdnn_tensor_dy, hipdnn_tensor_dw = build_conv_wrw_graph(
        hipdnn_handle,
        torch_tensor_x,
        torch_tensor_dy,
        [pad_h, pad_w],
        [stride_h, stride_w],
        [dil_h, dil_w],
        hipdnn_data_type,
    )
    torch_tensor_dw = torch.empty(hipdnn_tensor_dw.get_dim(), dtype=torch_data_type, device="cuda")
    variant_pack = {
        hipdnn_tensor_x: torch_tensor_x.data_ptr(),
        hipdnn_tensor_dy: torch_tensor_dy.data_ptr(),
        hipdnn_tensor_dw: torch_tensor_dw.data_ptr(),
    }
    workspace = torch.empty(graph.get_workspace_size(), dtype=torch.uint8, device="cuda")

    graph.exec(variant_pack=variant_pack, workspace=workspace.data_ptr())
    print("Convolution wrw graph execution complete.")