基于python3.9的pytorth-gpu安装环境简介
一、conda 安装
nvidia-smi
nvcc -V
conda env list
conda list
nvidia-smi
nvcc -V
conda env list
conda list
conda安装
下载anaconda 2022.10 window-x86-x64.exe
https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=A
设置系统Path
path+=
D:\ProgramData\Anaconda3
D:\ProgramData\Anaconda3\Scripts
D:\ProgramData\Anaconda3\Library\bin
D:\ProgramData\Anaconda3\Library\mingw-w64\bin
cuda安装
安装cuda和cudnn环境前：先要安装和升级显卡驱动程序
注意查看cuda或cudnn开发者库有没有安装，其显著特征就是：
有没有目录路径 C:\Program Files\NVIDIA GPU Computing Toolkit
cuda下载地址：https://developer.nvidia.com/cuda-toolkit-archive
我选择的是版本11.7版本
nvcc -V #测试是否安装成功
CUDNN的安装
下载地址：
https://developer.nvidia.com/rdp/cudnn-download
必须注册后下载cuda11版本 8.9的window版本
cudnn-windows-x86_64-8.9.4.25_cuda11-archive.zip
(就是将解压后得到的的bin ，include 和lib文件夹分别复制到cuda安装路径下与cuda的bin ，include 和lib文件夹合并)
国内镜像源配置
通过win+R输入cmd进入命令行，通过输入下面命令配置为清华源：
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
二、pytorch-gpu 安装
创建虚拟环境pytorch-gpu
conda create -n pytorch-gpu python=3.9
conda env list
conda create -n pytorch-gpu python=3.9
conda env list
C:\Users\Administrator>conda env list
# conda environments:
#
base D:\ProgramData\Anaconda3
pytorch-gpu D:\ProgramData\Anaconda3\envs\pytorch-gpu
D:\arcgispro_ai\arcgispro-py3-clone332
D:\arcgispro_ai\arcgispro-py3-clone342
D:\cwgis_AI\xinference
即创建和安装到目录：D:\ProgramData\Anaconda3\envs\pytorch-gpu
全局名称为：pytorch-gpu
python 3.9版本
python
python
Python 3.9.22 | packaged by conda-forge | (main, Apr 14 2025, 23:26:18) [MSC v.1929 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
激活（进入）虚拟环境
conda activate pytorch-gpu
conda activate pytorch-gpu
退出环境
codna deactivate
codna deactivate
首先，去pytorch官网
https://pytorch.org/
自动检测 CUDA 版本（推荐）：
访问 PyTorch 官网安装命令生成器，选择：
复制生成的命令（示例）：
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
手动指定 CUDA 版本（如已安装 CUDA 11.7）
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117
自动下载的版本是pytorch2.0.1+cu117+cp39：
https://download.pytorch.org/whl/cu117/torch-2.0.1%2Bcu117-cp39-cp39-win_amd64.whl
conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch
conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch
安装 CPU 版本（无显卡/临时使用）
conda install pytorch torchvision torchaudio cpuonly -c pytorch
conda install pytorch torchvision torchaudio cpuonly -c pytorch
vscode环境中验证安装环境：
testInstallCudaPytortch.py
import torch
# 检查版本
print(torch.__version__) # 输出版本，如 2.1.0
# 检查 GPU 支持（NVIDIA）
if torch.cuda.is_available():
print(f"GPU 型号: {torch.cuda.get_device_name(0)}")
print(f"CUDA 版本: {torch.version.cuda}")
# 测试 GPU 计算
x = torch.tensor([1.0]).cuda()
print(x) # 输出 tensor([1.0], device='cuda:0')
else:
print("未检测到 GPU，使用 CPU 版本")
# Apple Silicon 专用检查
if hasattr(torch.backends, "mps") and torch.backends.mps.is_available():
print("MPS 加速已启用")
y = torch.tensor([1.0], device="mps")
print(y)
testInstallCudaPytorchALL.py
import torch
import platform
def get_system_info():
return {
"系统": platform.system(),
"Python 版本": platform.python_version(),
"PyTorch 版本": torch.__version__,
"CUDA 可用": torch.cuda.is_available(),
"CUDA 版本": torch.version.cuda,
"MPS 可用": hasattr(torch.backends, "mps") and torch.backends.mps.is_available(),
"显卡信息": torch.cuda.get_device_name(0) if torch.cuda.is_available() else "无"
}
if __name__ == "__main__":
info = get_system_info()
for k, v in info.items():
print(f"{k}: {v}")
执行验证结果为：
PS D:\cwgis_python> & D:/ProgramData/Anaconda3/envs/pytorch-gpu/python.exe d:/cwgis_python/testInstallCudaPytortch.py
2.0.1+cu117
GPU 型号: NVIDIA GeForce GTX 1080
CUDA 版本: 11.7
tensor([1.], device='cuda:0')
PS D:\cwgis_python> & D:/ProgramData/Anaconda3/envs/pytorch-gpu/python.exe
d:/cwgis_python/testInstallCudaPytorchALL.py
系统: Windows
Python 版本: 3.9.22
PyTorch 版本: 2.0.1+cu117
CUDA 可用: True
CUDA 版本: 11.7
MPS 可用: False
显卡信息: NVIDIA GeForce GTX 1080
PS D:\cwgis_python>
vscode执行py文件需要设置vscode python执行环境
快捷键：Ctrl+Shift+P
选择python select interpreter
选择下面路径的pytorch-gpu版本
D:/ProgramData/Anaconda3/envs/pytorch-gpu/python.exe
(pytorch-gpu) C:\Users\Administrator>nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Jun__8_16:59:34_Pacific_Daylight_Time_2022
Cuda compilation tools, release 11.7, V11.7.99
Build cuda_11.7.r11.7/compiler.31442593_0
(pytorch-gpu) C:\Users\Administrator>nvidia-smi
Tue Apr 29 11:11:25 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 572.70 Driver Version: 572.70 CUDA Version: 12.8 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Driver-Model | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce GTX 1080 WDDM | 00000000:01:00.0 On | N/A |
| 37% 31C P8 11W / 250W | 875MiB / 8192MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 2380 C+G ...ntrolPanel\SystemSettings.exe N/A |
| 0 N/A N/A 2936 C+G ...ms\Microsoft VS Code\Code.exe N/A |
| 0 N/A N/A 9756 C+G ...t\Edge\Application\msedge.exe N/A |
| 0 N/A N/A 10440 C+G C:\Windows\explorer.exe N/A |
| 0 N/A N/A 10472 C+G ...indows\System32\ShellHost.exe N/A |
| 0 N/A N/A 11404 C+G ..._cw5n1h2txyewy\SearchHost.exe N/A |
| 0 N/A N/A 11428 C+G ...y\StartMenuExperienceHost.exe N/A |
| 0 N/A N/A 12392 C+G ...5n1h2txyewy\TextInputHost.exe N/A |
| 0 N/A N/A 13244 C+G ...yb3d8bbwe\Notepad\Notepad.exe N/A |
| 0 N/A N/A 13964 C+G ...e Experience\NVIDIA Share.exe N/A |
| 0 N/A N/A 14092 C+G ....0.3179.98\msedgewebview2.exe N/A |
| 0 N/A N/A 14816 C+G ...crosoft OneDrive\OneDrive.exe N/A |
| 0 N/A N/A 14948 C+G ...ffice6\promecefpluginhost.exe N/A |
| 0 N/A N/A 16012 C+G ...em32\ApplicationFrameHost.exe N/A |
| 0 N/A N/A 17252 C+G ...xyewy\ShellExperienceHost.exe N/A |
+-----------------------------------------------------------------------------------------+
(pytorch-gpu) C:\Users\Administrator>
本blog地址：https://blog.csdn.net/hsg77
扫一扫
4284
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
