how do I check If I install CUDA and CUDNN successfully?

I Flash TX2 with JetPack3.1.After it I tested CUDA example successfully!

And I typed nvcc -V on the terminal.it lists:

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2016 NVIDIA Corporation
Built on Mon_Mar_20_17:07:33_CDT_2017
Cuda compilation tools, release 8.0, V8.0.72

Then I typed: nvidia-smi,the result is:
bash: nvidia-smi: command not found
I typed:
sudo apt-get update
sudo apt-get install nvidia-smi

then it list:

Reading package lists... Done
Building dependency tree       
Reading state information... Done
Package nvidia-smi is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source
E: Package 'nvidia-smi' has no installation candidate

I do not know where is the problem.

Then I checked the folde /usr/local/cuda-8.0/include there is no cudnn.h
so do I successfully intsall CUDA and CuDNN

Hi,

nvidia-smi is not available on Jetson platform.
If you want to check GPU/CUDA status, please use deviceQuery sample:

$ /usr/local/cuda-8.0/bin/cuda-install-samples-8.0.sh .
$ cd NVIDIA_CUDA-8.0_Samples/1_Utilities/deviceQuery
$ make && ./deviceQuery

CUDA and cuDNN can be installed via JetPack3.1 directly.
cuDNN libraries is located at /usr/lib/aarch64-linux-gnu/

Thanks.

3 Likes

Hi,

Thanks for your answer.

$ /usr/local/cuda-8.0/bin/cuda-install-samples-8.0.sh
$ cd NVIDIA_CUDA-8.0_Samples/1_Utilities/deviceQuery
$ make && ./deviceQuery

When I ran the commands above, it cannot found any target file or folder at all.
But I do flash the Jetson TX2 with your Jetpack3.1

Any solution?

THX

Hi,

Please remember to specify the installation folder to the script.
For example:

$ /usr/local/cuda-8.0/bin/cuda-install-samples-8.0.sh <b>~</b>

This will install CUDA sample to the home directory.

Thanks.

2 Likes

Moreover, according to the article, you can also run ./bandwidthTest:

cd /usr/local/cuda-8.0/samples
sudo make
cd bin/x86_64/linux/release
sudo ./deviceQuery
sudo ./bandwidthTest
2 Likes

This is for cuda 9 but I think iā€™d work with cuda 8 too:

āžœ pwd
/usr/local/cuda-9.0/bin
āžœ ./nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2017 NVIDIA Corporation
Built on Sun_Nov_19_03:10:15_CST_2017
Cuda compilation tools, release 9.0, V9.0.252
āžœ
3 Likes

Device : Jetson Nano b01
JetPack Version : 4.4

Command used for checking CUDA version:

cat /usr/local/cuda/version.txt
7 Likes