Failed to install cuda to tx2

I am trying to build the tx2 by JetPack3.3.
But it can’t build CUDA sample while it show the error like below:

Depending on the component selection, please pay attention to the prompt in embedded terminal:

  • Additional user input may be required.
  • Follow on-screen instructions to continue.

So how can I fix the issue??

Can you ssh directly to the “nvidia” account from your host? For example, if the address is “”, then from the host:

ssh nvidia@

Beware that the first time you connect it’ll want you to confirm the other side of the connection is trusted. Saying yes on command line eliminates needing to do this on first try from a JetPack shell.

You may still need to manually enter the address and password…watch the console. Default password for “nvidia” is “nvidia” (if you are not behind a firewall or router stopping incoming outside connections, then you need to change passwords immediately after install).