当前位置:网站首页>Orin 安装CUDA环境
Orin 安装CUDA环境
2022-07-05 06:28:00 【恩来贺】
有两种方法可以安装CUDA环境
1.命令行方式
在刷机完成的Orin,执行如下命令:
a. sudo apt update
b. sudo apt upgrade
c. sudo apt install nvidia-jetpack -y
如果报错则,查看版本:/etc/apt/sources.list.d/nvidia-l4t-apt-source.list 文件中,目前最新的为34.1

改为

执行完毕即可安装,如果使用 l4t_for_tegra 最新的版本,通过flash.sh刷机之后不会报错。
a. CUDA 检查是否安装成功
nvcc -V
如果报错,需要把nvcc添加到环境变量。
出现如下则表示安装正确:
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Thu_Nov_11_23:44:05_PST_2021
Cuda compilation tools, release 11.4, V11.4.166
Build cuda_11.4.r11.4/compiler.30645359_0b.cuDNN
dpkg -l libcudnn8显示如下信息:
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-===================-============-======================
ii libcudnn8 8.3.2.49-1+cuda11.4 arm64 cuDNN runtime librariesc.TensorRT:
显示如下信息:
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-===================-============-=====================
ii tensorrt 8.4.0.11-1+cuda11.4 arm64 Meta package of TensorRTd.OpenCV:
显示如下信息:
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-===================-============-=======================
ii libopencv 4.5.4-8-g3e4c170df4 arm64 Open Computer Vision Library2. 使用SDKmanager 安装
在SDKmanager 下载所需要的文件
执行step1/2/3/4 即可,操作比较简单。
查看安装的版本:

TensorRT
TensorRT是一个高性能的深度学习推理(Inference)优化器,可以为深度学习应用提供低延迟、高吞吐率的部署推理。TensorRT可用于对超大规模数据中心、嵌入式平台或自动驾驶平台进行推理加速。 TensorRT现已能支持TensorFlow、Caffe、Mxnet、Pytorch等几乎所有的深度学习框架,将TensorRT和NVIDIA的GPU结合起来,能在几乎所有的框架中进行快速和高效的部署推理。
cuDNN
cuDNN是用于深度神经网络的GPU加速库。它强调性能、易用性和低内存开销。NVIDIA cuDNN可以集成到更高级别的机器学习框架中,如谷歌的Tensorflow、加州大学伯克利分校的流行caffe软件。简单的插入式设计可以让开发人员专注于设计和实现神经网络模型,而不是简单调整性能,同时还可以在GPU上实现高性能现代并行计算。
CUDA
CUDA(ComputeUnified Device Architecture),是显卡厂商NVIDIA推出的运算平台。 CUDA是一种由NVIDIA推出的通用并行计算架构,该架构使GPU能够解决复杂的计算问题。
CUDA与cuDNN的关系
CUDA看作是一个工作台,上面配有很多工具,如锤子、螺丝刀等。cuDNN是基于CUDA的深度学习GPU加速库,有了它才能在GPU上完成深度学习的计算。它就相当于工作的工具,比如它就是个扳手。但是CUDA这个工作台买来的时候,并没有送扳手。想要在CUDA上运行深度神经网络,就要安装cuDNN,就像你想要拧个螺帽就要把扳手买回来。这样才能使GPU进行深度神经网络的工作,工作速度相较CPU快很多。
边栏推荐
- Game theory acwing 894 Split Nim game
- Bit of MySQL_ OR、BIT_ Count function
- MPLS experiment
- 微信小程序路由再次跳转不触发onload
- Series of how MySQL works (VIII) 14 figures explain the atomicity of MySQL transactions and the principle of undo logging
- Redis-01. First meet redis
- cgroup_ memcg
- Client use of Argo CD installation
- Find the combination number acwing 889 01 sequence meeting conditions
- Database mysql all
猜你喜欢

Client use of Argo CD installation

MPLS experiment

Package webapp or H5 pages into apps

Some classic recursion problems

2. Addition and management of Oracle data files

vsCode创建自己的代码模板
![[Gaode map POI stepping pit] amap Placesearch cannot be used](/img/4c/55586ffcc2267c477a4532ab51a0c1.png)
[Gaode map POI stepping pit] amap Placesearch cannot be used

Genesis builds a new generation of credit system

Game theory acwing 892 Steps Nim game

Game theory acwing 891 Nim games
随机推荐
[QT] QT multithreading development qthread
Integer to 8-bit binary explanation (including positive and negative numbers) scope of application -127~+127
Install opencv -- CONDA to establish a virtual environment and add the kernel of this environment in jupyter
vim
[moviepy] unable to find a solution for exe
Error: “MountVolume.SetUp failed for volume pvc 故障处理
MQClientException: No route info of this topic: type_ topic
Sorting out the latest Android interview points in 2022 to help you easily win the offer - attached is the summary of Android intermediate and advanced interview questions in 2022
H5 module suspension drag effect
2022-5-the fourth week daily
Knapsack problem acwing 9 Group knapsack problem
Instruction execution time
Redis-02. Redis command
Package webapp or H5 pages into apps
Record of problems in ollvm compilation
Series of how MySQL works (VIII) 14 figures explain the atomicity of MySQL transactions and the principle of undo logging
Interval problem acwing 906 Interval grouping
5.Oracle-表空间
namespace
Edge calculation data sorting