Created by: OliverLPH

• 标题：DockerFile里 提供的TRT lib文件可能有误
• 版本、环境信息：    1）PaddlePaddle版本：    2）CPU：    3）GPU：CUDA8, cuDNN 7    4）系统环境：ubuntu1604
• 安装方式信息： 1）pip安装/docker安装 2）本地编译：请提供cmake命令，编译命令 3）docker编译：请提供docker镜像，编译命令             特殊环境请注明：如离线安装等
• 复现信息：
• 问题描述：TensorRT 库文件

在bce里提供的下载文件 https://paddlepaddledeps.bj.bcebos.com/TensorRT-4.0.1.6-ubuntu14.04.x86_64-gnu.cuda.8.0.cudnn7.0.tar.gz 中 NvInfer.h 貌似有变动。

以前

class IGpuAllocator
{
public:
    virtual ~IGpuAllocator() {}
    /**
     * a callback implemented by the application to handle acquisition of GPU memory
     *
     * \param size the size of the memory required
     * \param alignment the alignment of the memory required. Alignment will zero or be a power of 2 not greater than 256, and thus this allocator can be safely implemented with cudaMalloc/cudaFree. An alignment value of zero indicates any alignment is acceptable.
     * \param flags reserved for future use. In the current release, 0 will be passed.
     *
     * If an allocation request of size 0 is made, nullptr should be returned.
     *
     * If an allocation request cannot be satisfied, nullptr should be returned.
     */
    virtual void* allocate(uint64_t size, uint64_t alignment, uint32_t flags) = 0;

    /**
     * a callback implemented by the application to handle release of GPU memory.
     *
     * TensorRT may pass a nullptr to this function if it was previously returned by allocate()
     *
     * \param memory the acquired memory.
     */
    virtual void free(void* memory) = 0;
};

现在

class IGpuAllocator
{
public:
    //!
    //! A callback implemented by the application to handle acquisition of GPU memory.
    //!
    //! \param size The size of the memory required.
    //! \param alignment The alignment of the memory required. Alignment will zero or be a power of 2 not greater than 256, and thus this allocator can be safely implemented with cudaMalloc/cudaFree. An alignment value of zero indicates any alignment is acceptable.
    //! \param flags Reserved for future use. In the current release, 0 will be passed.
    //!
    //! If an allocation request of size 0 is made, nullptr should be returned.
    //!
    //! If an allocation request cannot be satisfied, nullptr should be returned.
    //!
    virtual void* allocate(uint64_t size, uint64_t alignment, uint32_t flags) = 0;

    //!
    //! A callback implemented by the application to handle release of GPU memory.
    //!
    //! TensorRT may pass a nullptr to this function if it was previously returned by allocate().
    //!
    //! \param memory The acquired memory.
    //!
    virtual void free(void* memory) = 0;
};

现在的文件里面缺乏析构函数，编译的时候会报错。