【Deepin 20系统+Tensorflow 2】Linux系统解决Could not load dynamic library ‘libcudart.so.10.0‘

港控/mmm° 2022-11-06 07:58 415阅读 0赞

# 问题 #

环境

> Deepin 20系统： Debian OS  
> NVIDIIA 驱动: NVIDIA-SMI 455.45.01 Driver Version: 455.45.01  
> nvidia-smi 显示cuda版本: CUDA Version: 11.1  
> 安装的Cuda版本：11.1  
> Tensorflow2.0

在使用tensorflow2跑程序时，GPU没有利用起来，通过代码查看GPU使用情况，输出日志如下

import tensorflow as tf
    print(tf.test.is_gpu_available())

输出日志

2021-03-08 19:51:11.747354: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
    2021-03-08 19:51:11.774909: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2899885000 Hz
    2021-03-08 19:51:11.775871: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55d501378430 executing computations on platform Host. Devices:
    2021-03-08 19:51:11.775926: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Host, Default Version
    2021-03-08 19:51:11.779414: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
    2021-03-08 19:51:11.881523: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
    2021-03-08 19:51:11.882130: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55d50140f0f0 executing computations on platform CUDA. Devices:
    2021-03-08 19:51:11.882152: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): GeForce RTX 2070, Compute Capability 7.5
    2021-03-08 19:51:11.882334: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
    2021-03-08 19:51:11.882678: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties: 
    name: GeForce RTX 2070 major: 7 minor: 5 memoryClockRate(GHz): 1.62
    pciBusID: 0000:01:00.0
    2021-03-08 19:51:11.882903: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudart.so.10.0'; dlerror: libcudart.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.882991: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcublas.so.10.0'; dlerror: libcublas.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.883060: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcufft.so.10.0'; dlerror: libcufft.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.883131: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcurand.so.10.0'; dlerror: libcurand.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.883202: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusolver.so.10.0'; dlerror: libcusolver.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.883269: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusparse.so.10.0'; dlerror: libcusparse.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.883334: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudnn.so.7'; dlerror: libcudnn.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/mgege007/cuda-11/lib64::/usr/local/cuda/extras/CUPTI/lib64
    2021-03-08 19:51:11.883344: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1641] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
    Skipping registering GPU devices...
    2021-03-08 19:51:11.883362: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:
    2021-03-08 19:51:11.883370: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165]      0 
    2021-03-08 19:51:11.883376: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0:   N 
    False

# 问题解决 #

[参考资料][Link 1]

> 日志说明：正在寻找CUDA 10.0。 但是我的CUDA版本是11.1，不能使用CUDA 11.1或CUDA 11.2替代它。 仅仅需要去官网下载CUDA10.0和对应的cudnn安装即可，不需要更改NVIDIA驱动。并且注意，根据自己的文件CUDA路径，进行相应的环境变量的配置

步骤  
（1）CUDA+CuDnn安装过程参考本人另一篇博客[Deepin 20系统安装CUDA+cudnn][Deepin 20_CUDA_cudnn]  
（2）更改CUDA环境变量参考本人另一篇博客[Deepin 20系统切换CUDA版本][Deepin 20_CUDA]

[Link 1]: https://stackoverflow.com/questions/64880546/tensorflow-could-not-load-dynamic-library-libcudart-so-10-0-on-ubuntu-18-04
[Deepin 20_CUDA_cudnn]: https://betterbench.blog.csdn.net/article/details/110133068
[Deepin 20_CUDA]: https://betterbench.blog.csdn.net/article/details/114500069