适用GPU (这个要找个通用的方法,根据显卡、windows系统,找到对应版本的cuda cunn pytorch)
安装 cuda cunn 和显卡型号 都要匹配上,不然会出现莫名其妙的问题,包括 pytorch也要匹配上。
注意事项:不同的算法平台,对cuda和cunn的版本需求都不一样,为避免发生乱七八糟的错误,需要保持版本的一致。例如pytorch 与 cuda cunn要保持一致
一、介绍
cuda : 简单地说,CUDA 是一个框架,允许你用 C/C++(或其他支持的语言)直接为 GPU 编写程序。例如,让代码能够使用GPUcunn:是一个专为深度学习而设计的库,旨在使神经网络操作在支持CUDA的NVIDIA GPU上得到加速。它是Torch深度学习框架的一个组件pytorch:PyTorch 是一个开源的深度学习框架paddlepaddle:PaddlePaddle(Parallel Distributed Deep Learning) 是由百度研发并开源的深度学习框架二、版本选型与安装
1、cuda选择并下载安装
1.1 输入CMD指令 nvidia-smi ,查看自己电脑的显卡型号。或者用其他方式
1.2. 找到显卡算力 和 对应架构 地址
1.3. 下载cuda 地址
点击对应各个版本,去看自己的想下载的版本支持的算力和架构
这里直接提供一个总结后的清单列表
Supported CUDA level of GPU and card.CUDA SDK 1.0 support for compute capability 1.0 – 1.1 (TeslaCUDA SDK 1.1 support for compute capability 1.0 – 1.1+x (Tesla)CUDA SDK 2.0 support for compute capability 1.0 – 1.1+x (Tesla)CUDA SDK 2.1 – 2.3.1 support for compute capability 1.0 – 1.3 (Tesla)CUDA SDK 3.0 – 3.1 support for compute capability 1.0 – 2.0 (Tesla, Fermi)CUDA SDK 3.2 support for compute capability 1.0 – 2.1 (Tesla, Fermi)CUDA SDK 4.0 – 4.2 support for compute capability 1.0 – 2.1+x (Tesla, Fermi, more).CUDA SDK 5.0 – 5.5 support for compute capability 1.0 – 3.5 (Tesla, Fermi, Kepler).CUDA SDK 6.0 support for compute capability 1.0 – 3.5 (Tesla, Fermi, Kepler).CUDA SDK 6.5 support for compute capability 1.1 – 5.x (Tesla, Fermi, Kepler, Maxwell). Last version with support for compute capability 1.x (Tesla).CUDA SDK 7.0 – 7.5 support for compute capability 2.0 – 5.x (Fermi, Kepler, Maxwell).CUDA SDK 8.0 support for compute capability 2.0 – 6.x (Fermi, Kepler, Maxwell, Pascal). Last version with support for compute capability 2.x (Fermi).CUDA SDK 9.0 – 9.2 support for compute capability 3.0 – 7.0 (Kepler, Maxwell, Pascal, Volta)CUDA SDK 10.0 – 10.2 support for compute capability 3.0 – 7.5 (Kepler, Maxwell, Pascal, Volta, Turing). Last version with support for compute capability 3.0 and 3.2 (Kepler in part). 10.2 is the last official release for macOS, as support will not be available for macOS in newer releases.CUDA SDK 11.0 support for compute capability 3.5 – 8.0 (Kepler (in part), Maxwell, Pascal, Volta, Turing, Ampere (in part)).CUDA SDK 11.1 – 11.4 support for compute capability 3.5 – 8.6 (Kepler (in part), Maxwell, Pascal, Volta, Turing, Ampere (in part)).CUDA SDK 11.5 – 11.7.1 support for compute capability 3.5 – 8.7 (Kepler (in part), Maxwell, Pascal, Volta, Turing, Ampere).CUDA SDK 11.8 support for compute capability 3.5 – 9.0 (Kepler (in part), Maxwell, Pascal, Volta, Turing, Ampere, Ada Lovelace, Hopper).CUDA SDK 12.0 support for compute capability 5.0 – 9.0 (Maxwell, Pascal, Volta, Turing, Ampere, Ada Lovelace, Hopper)
### 二、panddlepanddle版本选择与安装 cunn版本查找,版本对应地址
下载地址 需要有账号,直接用苹果ID账号登录也可以cunn安装
下载好了,把下载包解压。将里面的文件翻盖到cuda安装路径中。
(cunn的bin目录里面文件,全部放到 cuda安装路径中的bin目录。其余文件也类似)paddlepaddle-gpu安装
(1)paddlepaddle-gpu 使用 pip 安装后默认使用GPU参加计算
(2)paddlepaddle 安装后,默认使用CPU参加计算
(3)python -m pip install paddlepaddle-gpu==2.4.2.post117 -f .cn/whl/windows/mkl/avx/stable.html
(4)cuda 117 paddle版本 2.4.2 最新的 2.5.1需要7.0以上算力的架构,我1050TI显卡(算力6.1)暂时不满足
三、pytorch版本选择与安装
依照 paddlepaddle
错误
错误一
Could not locate zlibwapi.dll. Please make sure it is in your library path!Process finished with exit code -1073740791 (0xC0000409)解决:安装 zlibwapi.dll 数据压缩的库zlibwapi.lib文件放到 C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.5/libzlibwapi.dll文件放到 C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.5/bin原文链接:/qq_40280673/article/details/132229908
错误二
The GPU architecture in your current machine is Pascal, which is not compatible with Paddle installation with arch: 70 75 80 86 , it is recommended to install the corresponding wheel package according to the installation information on the official Paddle website paddlepaddle-gpu版本是2.5.1 ,需要显卡算力 7.0以上,我的显卡算力只有6.1,所以不能使用
如果觉得《如何用windows使用GPU进行人工智能》对你有帮助,请点赞、收藏,并留下你的观点哦!