好几次遇到问为什么安装的 tensorflow 不能调用GPU，之前搞定过几次，前两天又有人问，又捣鼓了很久才搞定，这里简单记录一下我遇到的问题，以及解决方案。
一、安装方法
（一）安装并更新 conda
1.安装 conda
安装 conda 很重要，使用 pip 安装 tensorflow-gpu 太多问题了（这里默认已经安装了conda）。
2.更新 conda
conda update -n base -c defaults conda --repodata-fn=repodata.json之前根据百度，都是执行：
conda update -n base -c defaults conda然后，首先该命令无法更新到最新的 conda；其次，我们在使用 conda -V 查看版本时，conda 版本显示错误。
该解决方案来自于 GitHub：I got update warning message but unable to update · Issue #12519 · conda/conda · GitHub
https://github.com/conda/conda/issues/12519
将 conda 的 base 更新到最新，我觉得原因是能够同步最新的包依赖关系，过时的版本可能导致依赖出问题。
（二）创建环境
1.创建环境
conda create -n TensorFlow2.4 python=3.9当然，这里可以根据自己的 CUDA 版本选择对应的 tensorflow 版本，我的 CUDA 版本为 11.3 ：
(Tensorflow2.4) name@eclab:~$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Sun_Mar_21_19:15:46_PDT_2021
Cuda compilation tools, release 11.3, V11.3.58
Build cuda_11.3.r11.3/compiler.29745058_0不显示的话，可以自行进一步搜索为什么 nvcc -V 不显示：
Ubuntu20.04LTS系统CUDA已经安装但nvcc -V显示command not found_nvcc -v 提示未找到命令_AISecurity盐究员的博客-CSDN博客安装了NVIDIA驱动程序，同时也安装了CUDA，但使用nvcc -V使用nvcc -V命令可以查看CUDA的版本，如下所示为正常的输入、输出内容，可以看出通过nvcc -V命令，可以看到目前所使用的CUDA版本。_nvcc -v 提示未找到命令https://blog.csdn.net/m0_38068876/article/details/127836484 注 .bashrc 文件添加环境变量时，需要根据 /usr/local/ 下的 cuda实际情况进行修改，这里展示我的情况：
(Tensorflow2.4) name@eclab:~$ cd /usr/local/
(Tensorflow2.4) name@eclab:/usr/local$ ls
bin cuda-11.3 games lib sbin src
cuda etc include man share sunlogin这里有 cuda 软链接，链接到 cuda-11.3，所以建议使用下面命令：
# cuda-11.3
export PATH=/usr/local/cuda-11.3/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-11.3/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
（三）安装 tensorflow-gpu
1.安装
激活环境：
conda activate TensorFlow2.4安装 tensorflow-gpu：
conda install tensorflow-gpu注：不要使用pip安装！不要使用pip安装！不要使用pip安装！
这里没有选择 tensorflow-gpu 版本，conda 自动下载了 tensorflow-gpu==2.4.1 （版本对应可以查看 Build from source | TensorFlow）。
2.测试
执行如下两个命令即可：
(Tensorflow2.4) name@eclab:/usr/local$ python
Python 3.9.17 (main, Jul 5 2023, 20:41:20)
[GCC 11.2.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
2023-07-10 10:22:16.571135: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.10.1
>>> tf.config.list_physical_devices('GPU')
2023-07-10 10:22:27.565493: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set
2023-07-10 10:22:27.567453: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcuda.so.1
2023-07-10 10:22:27.611185: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 0 with properties:
pciBusID: 0000:02:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:22:27.612680: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 1 with properties:
pciBusID: 0000:03:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:22:27.613857: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 2 with properties:
pciBusID: 0000:82:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:22:27.614783: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 3 with properties:
pciBusID: 0000:83:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:22:27.614821: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.10.1
2023-07-10 10:22:27.617316: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcublas.so.10
2023-07-10 10:22:27.617370: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcublasLt.so.10
2023-07-10 10:22:27.619509: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcufft.so.10
2023-07-10 10:22:27.619882: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcurand.so.10
2023-07-10 10:22:27.622449: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcusolver.so.10
2023-07-10 10:22:27.623913: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcusparse.so.10
2023-07-10 10:22:27.629319: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudnn.so.7
2023-07-10 10:22:27.644606: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1862] Adding visible gpu devices: 0, 1, 2, 3
[PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'), PhysicalDevice(name='/physical_device:GPU:1', device_type='GPU'), PhysicalDevice(name='/physical_device:GPU:2', device_type='GPU'), PhysicalDevice(name='/physical_device:GPU:3', device_type='GPU')]（四）使用 pip 安装的问题
下面演示使用 pip 安装的话存在的问题。
(base) name@eclab:~$ conda create -n Tensorflow-err python=3.9
(Tensorflow-err) name@eclab:~$ pip install tensorflow-gpu==2.4.1
ERROR: Could not find a version that satisfies the requirement tensorflow-gpu==2.4.1 (from versions: 2.5.0, 2.5.1, 2.5.2, 2.5.3, 2.6.0, 2.6.1, 2.6.2, 2.6.3, 2.6.4, 2.6.5, 2.7.0rc0, 2.7.0rc1, 2.7.0, 2.7.1, 2.7.2, 2.7.3, 2.7.4, 2.8.0rc0, 2.8.0rc1, 2.8.0, 2.8.1, 2.8.2, 2.8.3, 2.8.4, 2.9.0rc0, 2.9.0rc1, 2.9.0rc2, 2.9.0, 2.9.1, 2.9.2, 2.9.3, 2.10.0rc0, 2.10.0rc1, 2.10.0rc2, 2.10.0rc3, 2.10.0, 2.10.1, 2.11.0rc0, 2.11.0rc1, 2.11.0rc2, 2.11.0, 2.12.0)
ERROR: No matching distribution found for tensorflow-gpu==2.4.1首先安装不了2.4.1，根据提示，选择安装2.5.0；
pip install tensorflow-gpu==2.5使用步骤（三）中的 2.测试方法：
(Tensorflow-err) name@eclab:~$ python
Python 3.9.17 (main, Jul 5 2023, 20:41:20)
[GCC 11.2.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
2023-07-10 10:38:37.238756: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
>>> tf.config.list_physical_devices('GPU')
2023-07-10 10:38:40.413250: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcuda.so.1
2023-07-10 10:38:40.456066: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 0 with properties:
pciBusID: 0000:02:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:38:40.457549: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 1 with properties:
pciBusID: 0000:03:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:38:40.458707: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 2 with properties:
pciBusID: 0000:82:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:38:40.459651: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 3 with properties:
pciBusID: 0000:83:00.0 name: NVIDIA TITAN X (Pascal) computeCapability: 6.1
coreClock: 1.531GHz coreCount: 28 deviceMemorySize: 11.91GiB deviceMemoryBandwidth: 447.48GiB/s
2023-07-10 10:38:40.459700: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
2023-07-10 10:38:40.464266: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublas.so.11
2023-07-10 10:38:40.464333: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublasLt.so.11
2023-07-10 10:38:40.465775: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcufft.so.10
2023-07-10 10:38:40.466117: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcurand.so.10
2023-07-10 10:38:40.467045: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcusolver.so.11
2023-07-10 10:38:40.468303: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcusparse.so.11
2023-07-10 10:38:40.468555: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-11.3/lib64
2023-07-10 10:38:40.468578: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1766] 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...
[]
错误内容：
tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-11.3/lib64
