一整晚上安装失败,早上起来换源瞬间装完
开始验证
首先
nvidia-smi
卡都在呢,虽然CUDA Toolkit官方的支持文档里面没写Kepler架构,但是能用
接下来验证pytorch
首先来一句我记住的第一个conda命令
conda env list
接着第二句,,
conda activate [name]
验证pytorch是否安装成功:验证pytorch是否安装成功(linux) - 简书
nvcc not found 是cudatoolkit没有配置环境变量导致的
开始配置环境变量:cuda已经安装,但nvcc找不到 - 知乎
把他的10.1改成11.1,再
source ~/.bashrc
还有一个py文件,学长让我们感受深度学习的GPU加速能力
import torch
import time
a = torch.randn(400,600,600).to("cuda:0")
b = torch.randn(400,600,600).to("cuda:0")
start_time = time.time()
for i in range(1,1000):
c = a*b
end_time = time.time()
print("CUDA time: ",(end_time-start_time))
a = torch.randn(400,600,600)
b = torch.randn(400,600,600)
start_time = time.time()
for i in range(1,1000):
c = a*b
end_time = time.time()
print("CPU time: ",(end_time-start_time))
跑一下
加速还是很明显的,一共快了
python
Python 3.8.8 (default, Apr 13 2021, 19:58:26)
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> 114.60520601272583/0.015193939208984375
7542.823782324881
>>>
这么多倍(上面是1030跑的下面是K80跑的
至此PyTorch已验证配置完毕