Can't intall CUDA successfully with SDKmanager

There are the errors.
22:53:47 ERROR : CUDA Toolkit for L4T : command scp -q -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no /tmp/ hld@; ssh -t -q -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no hld@ “~/ && rm -f ~/” finished with error
22:53:47 ERROR : CUDA Toolkit for L4T : command terminated with error

please help me, I have tried 3 times.thank you!

Hi 582693898,

Could you try using eth0 ip address instead of usb device ip (, then install SDK components again. Thanks!

I used the eth0 ip, appearing:Cannot connect to the device via SSH.

Hi 582693898,

Can you connect to tx2 via SSH from host machine? please confirm. Thanks!

Hi carolyuu, my host machine’s IP is, TX2’s IP is,they are connected to the same network router,wo insert the TX2’s ip in the SDKmanager space,

Hi 582693898,

Can you please zip and upload the ~/.nvsdkm directory for more investigation?

sorry, I do not know how to upload, I have not found where to upload

If you hover your mouse over the quote icon of one of your existing posts, then other icons will show up. The paper clip icon is for attaching.

Hi 582693898,

Please check attached picture to upload the file.
External Media

Thank all of you for help!

But it still can’t be installed successfully,I have upload some screenshots.

In addition,where can I find the installed software after I install it

Hi 582693898,

The JetPack packages download on: /home/username>nvidia/nvidia_sdk/
The setup files downloads folder, expand the Download & Install Options drop-down menu, then select the path you wish to use:
External Media

Also, the CUDA-related items are typically in “/usr/local/”.

Thank you carolyuu and linuxdev .I have installed successfully, I flashed OS first and then installed target components, The screen showed that the installation is successful. So how i can check i have installed these components completely,is there any demo? Where to run it?

Hi 582693898,

You can check image version with below command:

$ head -1 /etc/nv_tegra_release

Check your sdk components:

  • CUDA: /usr/local
  • CUDNN, MMAPI, TensorRT: /usr/src/