RuntimeError: (PreconditionNotMet) Cannot load cudnn shared library. Cannot invoke method cudnnGetVersion. [Hint: cudnn_dso_handle should not be null.] (at /paddle/paddle/phi/backends/dynload/cudnn.cc:64)

[elontusk123]

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

• 系统环境/System Environment： Linux 20.0.4
• 版本号/Version：Paddle： PaddleOCR： 问题相关组件/Related components： gpu: NVIDIA-RTX A2000 12GB, nvcc 10.1, driver version - 535.161.07
• 运行指令/Command Code：result = ocr.ocr(img_path, cls=True)
• 完整报错/Complete Error Message：>>> img_path = '/home/usr/Documents/backend/image/a_picked.jpg'

result = ocr.ocr(img_path, cls=True)
Traceback (most recent call last):
File "", line 1, in
File "/home/usr/anaconda3/envs/env/lib/python3.8/site-packages/paddleocr/paddleocr.py", line 661, in ocr
dt_boxes, rec_res, _ = self.call(img, cls)
File "/home/usr/anaconda3/envs/env/lib/python3.8/site-packages/paddleocr/tools/infer/predict_system.py", line 76, in call
dt_boxes, elapse = self.text_detector(img)
File "/home/usr/anaconda3/envs/env/lib/python3.8/site-packages/paddleocr/tools/infer/predict_det.py", line 244, in call
self.input_tensor.copy_from_cpu(img)
File "/home/usr/anaconda3/envs/env/lib/python3.8/site-packages/paddle/inference/wrapper.py", line 52, in tensor_copy_from_cpu
self._copy_from_cpu_bind(data)
RuntimeError: (PreconditionNotMet) Cannot load cudnn shared library. Cannot invoke method cudnnGetVersion.
[Hint: cudnn_dso_handle should not be null.] (at /paddle/paddle/phi/backends/dynload/cudnn.cc:64)

请尽量不要包含图片在问题中/Please try to not include the image in the issue.

[a31413510]

请问你这是在运行什么命令或者做什么操作的时候报错的

[Shubham654]

Getting same error when calling the model in concurrently also when the Memory gets consumed and not released.

[plmsmile]

same cuda 12.4 [Hint: cudnn_dso_handle should not be null.] (at /paddle/paddle/phi/backends/dynload/cudnn.cc:64)

[kevenleung]

对系统cuda11.8的支持非常好，一次性就能安装成功，对新的cuda12.4支持不友好，各种试错，最终大概步骤如下，不一定保真，哈哈哈

1.CUDA11.8 包含 cuDNN 动态链接库的 PaddlePaddle，需要先使用如下命令将 wheel 包下载到本地，再使用python3 -m pip install [name].whl本地安装（[name]为 wheel 包名称）：
https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/install/pip/linux-pip.html

2.确保环境中安装有以下包：（版本号仅供参考）
nvidia-cublas-cu12 12.1.3.1 pypi_0 pypi
nvidia-cuda-cupti-cu12 12.1.105 pypi_0 pypi
nvidia-cuda-nvrtc-cu12 12.1.105 pypi_0 pypi
nvidia-cuda-runtime-cu12 12.1.105 pypi_0 pypi
nvidia-cudnn-cu12 8.9.2.26 pypi_0 pypi
nvidia-cufft-cu12 11.0.2.54 pypi_0 pypi
nvidia-curand-cu12 10.3.2.106 pypi_0 pypi
nvidia-cusolver-cu12 11.4.5.107 pypi_0 pypi
nvidia-cusparse-cu12 12.1.0.106 pypi_0 pypi
nvidia-nccl-cu12 2.20.5 pypi_0 pypi
nvidia-nvjitlink-cu12 12.4.127 pypi_0 pypi
nvidia-nvtx-cu12 12.1.105 pypi_0 pypi

3.如安装有paddleocr，请卸载（好像用不着）

4.如果已经安装paddlepaddle-gpu，重装为2.6.0的版本

5.设置以下缺失的动态库：
libcublas.so
libcudnn.so
libcudnn_cnn_infer.so.8
libcudnn_ops_infer.so.8

※ 确保系统环境变量设置为：
cuda-12.4为对象的cuda版本
export PATH=/usr/local/cuda/bin:/usr/local/cuda-12.4/bin:/usr/local/cuda-12.4/compute-sanitizer${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-12.4/lib64{$LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/usr/lib:$LD_LIBRARY_PATH

※ 根据上面的四个动态库，依次执行下面操作步骤：
① 定位系统中已存在的动态库
locate libcublas.so
locate libcudnn.so
locate libcudnn_cnn_infer.so.8
locate libcudnn_ops_infer.so.8
② 软连接到/usr/lib/路径下
sudo ln -s 源路径xxx /usr/lib/libcublas.so
sudo ln -s 源路径xxx /usr/lib/libcudnn.so
sudo ln -s 源路径xxx /usr/lib/libcudnn_cnn_infer.so.8
sudo ln -s 源路径xxx /usr/lib/libcudnn_ops_infer.so.8

6.进入项目，开启日志：
注释config.disable_glog_info()即可
查看报错日志，并调试

希望以上操作能帮上您！

[Win-commit]

Cuda 11.8 有一样的问题，

[Harry-KIT]

installing the following ones helped me to solve:

sudo apt-get install libcudnn8
sudo apt-get install libcudnn8-dev