目录
条件
(2024-12-25)
- 操作系统:Windows 11 专业版 24H2
- GPU:Geforce RTX 2080 Ti
- CPU:AMD Ryzen 5 9600X 6-Core Processor 3.90 GHz
预操作:
- 删除原安装的CUDA和cuDNN(未安装可忽略)
- 安装Visual Studio最新版(默认)
- 安装Anaconda最新版(默认)
环境创建:Python3.9
使用
Anaconda Prompt
命令提示符
conda create -n tf_gpu python=3.9
(base) C:\Users\jing>conda create -n tf_gpu python=3.9
Channels:
- defaults
Platform: win-64
Collecting package metadata (repodata.json): done
Solving environment: done
## Package Plan ##
environment location: D:\anaconda3\envs\tf_gpu
added / updated specs:
- python=3.9
The following NEW packages will be INSTALLED:
ca-certificates pkgs/main/win-64::ca-certificates-2024.11.26-haa95532_0
openssl pkgs/main/win-64::openssl-3.0.15-h827c3e9_0
pip pkgs/main/win-64::pip-24.2-py39haa95532_0
python pkgs/main/win-64::python-3.9.21-h8205438_1
setuptools pkgs/main/win-64::setuptools-75.1.0-py39haa95532_0
sqlite pkgs/main/win-64::sqlite-3.45.3-h2bbff1b_0
tzdata pkgs/main/noarch::tzdata-2024b-h04d1e81_0
vc pkgs/main/win-64::vc-14.40-haa95532_2
vs2015_runtime pkgs/main/win-64::vs2015_runtime-14.42.34433-h9531ae6_2
wheel pkgs/main/win-64::wheel-0.44.0-py39haa95532_0
Proceed ([y]/n)? y
CUDA与cuDNN安装:指定版本
https://www.tensorflow.org/install/pip#windows-native
conda activate tf_gpu
conda install -c conda-forge cudatoolkit=11.2 cudnn=8.1.0
(tf_gpu) C:\Users\jing>conda list
# packages in environment at D:\anaconda3\envs\tf_gpu:
#
# Name Version Build Channel
ca-certificates 2024.12.14 h56e8100_0 conda-forge
cudatoolkit 11.2.2 h7d7167e_13 conda-forge
cudnn 8.1.0.77 h3e0f4f4_0 conda-forge
openssl 3.4.0 h2466b09_0 conda-forge
pip 24.2 py39haa95532_0
python 3.9.21 h8205438_1
setuptools 75.1.0 py39haa95532_0
sqlite 3.45.3 h2bbff1b_0
tzdata 2024b h04d1e81_0
ucrt 10.0.22621.0 h57928b3_1 conda-forge
vc 14.40 haa95532_2
vc14_runtime 14.42.34433 he29a5d6_23 conda-forge
vs2015_runtime 14.42.34433 hdffcdeb_23 conda-forge
wheel 0.44.0 py39haa95532_0
TensorFlow安装:<2.11
Anything above 2.10 is not supported on the GPU on Windows Native
python -m pip install "tensorflow<2.11"
(tf_gpu) C:\Users\jing>conda list
# packages in environment at D:\anaconda3\envs\tf_gpu:
#
# Name Version Build Channel
absl-py 2.1.0 pypi_0 pypi
astunparse 1.6.3 pypi_0 pypi
ca-certificates 2024.12.14 h56e8100_0 conda-forge
cachetools 5.5.0 pypi_0 pypi
certifi 2024.12.14 pypi_0 pypi
charset-normalizer 3.4.1 pypi_0 pypi
cudatoolkit 11.2.2 h7d7167e_13 conda-forge
cudnn 8.1.0.77 h3e0f4f4_0 conda-forge
flatbuffers 24.12.23 pypi_0 pypi
gast 0.4.0 pypi_0 pypi
google-auth 2.37.0 pypi_0 pypi
google-auth-oauthlib 0.4.6 pypi_0 pypi
google-pasta 0.2.0 pypi_0 pypi
grpcio 1.68.1 pypi_0 pypi
h5py 3.12.1 pypi_0 pypi
idna 3.10 pypi_0 pypi
importlib-metadata 8.5.0 pypi_0 pypi
keras 2.10.0 pypi_0 pypi
keras-preprocessing 1.1.2 pypi_0 pypi
libclang 18.1.1 pypi_0 pypi
markdown 3.7 pypi_0 pypi
markupsafe 3.0.2 pypi_0 pypi
numpy 2.0.2 pypi_0 pypi
oauthlib 3.2.2 pypi_0 pypi
openssl 3.4.0 h2466b09_0 conda-forge
opt-einsum 3.4.0 pypi_0 pypi
packaging 24.2 pypi_0 pypi
pip 24.2 py39haa95532_0
protobuf 3.19.6 pypi_0 pypi
pyasn1 0.6.1 pypi_0 pypi
pyasn1-modules 0.4.1 pypi_0 pypi
python 3.9.21 h8205438_1
requests 2.32.3 pypi_0 pypi
requests-oauthlib 2.0.0 pypi_0 pypi
rsa 4.9 pypi_0 pypi
setuptools 75.1.0 py39haa95532_0
six 1.17.0 pypi_0 pypi
sqlite 3.45.3 h2bbff1b_0
tensorboard 2.10.1 pypi_0 pypi
tensorboard-data-server 0.6.1 pypi_0 pypi
tensorboard-plugin-wit 1.8.1 pypi_0 pypi
tensorflow 2.10.1 pypi_0 pypi
tensorflow-estimator 2.10.0 pypi_0 pypi
tensorflow-io-gcs-filesystem 0.31.0 pypi_0 pypi
termcolor 2.5.0 pypi_0 pypi
typing-extensions 4.12.2 pypi_0 pypi
tzdata 2024b h04d1e81_0
ucrt 10.0.22621.0 h57928b3_1 conda-forge
urllib3 2.3.0 pypi_0 pypi
vc 14.40 haa95532_2
vc14_runtime 14.42.34433 he29a5d6_23 conda-forge
vs2015_runtime 14.42.34433 hdffcdeb_23 conda-forge
werkzeug 3.1.3 pypi_0 pypi
wheel 0.44.0 py39haa95532_0
wrapt 1.17.0 pypi_0 pypi
zipp 3.21.0 pypi_0 pypi
验证安装:出错
Verify the installation:
python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
(tf_gpu) C:\Users\jing>python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.0.2 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.
If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time t