Method run_op

src/gpu/cl/kernels/ClCol2ImKernel.cpp:171–200

}

void ClCol2ImKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
{
    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
    ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(IClKernel::window(), window);

    bool is_collapsed     = false;
    bool is_collapsed_out = false;

    auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC));
    auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST));

    Window out_window;
    out_window.use_tensor_dimensions(dst->info()->tensor_shape());

    Window collapsed     = window.collapse_if_possible(ICLKernel::window(), Window::DimZ, &is_collapsed);
    Window collapsed_out = out_window.collapse_if_possible(out_window, 3, &is_collapsed_out);

    ARM_COMPUTE_ERROR_ON(is_collapsed != is_collapsed_out);

    Window slice     = collapsed.first_slice_window_3D();
    Window slice_out = collapsed_out.first_slice_window_4D();
    do
    {
        // Set inputs
        unsigned int idx = 0;
        add_3D_tensor_argument(idx, src, slice);
        add_4D_tensor_argument(idx, dst, slice_out);
        enqueue(queue, *this, slice, lws_hint());
    } while (collapsed.slide_window_slice_3D(slice) && collapsed_out.slide_window_slice_4D(slice_out));
}
} // namespace kernels
} // namespace opencl
} // namespace arm_compute