Jetson TX2 Developer Kit

Hello,

I just received recently a Jetson TX2 Developer Kit, but the Quick Start Guide provided is very brief, could I have detailed Startup Guide/Installation Manual to get started, anyone that could advise me the links where I could download these documentations will be most helpful.

Thanks and Best Regards
Hoe Min

I think anything you could possibly want can be found here:
https://developer.nvidia.com/embedded/downloads

I guess, have you managed to install L4T (Linux For Tegra). Do you have a Ubuntu environment when you start the TX2 Devkit up, or just a black screen with some white text? If possible, could be a little bit more specific about your current situation and what you’re trying to achieve?

Thanks,

David.

See the JetPack-L4T Jetson Developer Kit User Guide for the walkthrough of setting up and using your TX2. Upon recieving the devkit, it is recommended to update your Jetson with the latest JetPack-L4T board software (currently JetPack 3.1 / Linux4Tegra R28.1).

Dear David and Dusty,

Thanks for your email.

I have wired up the TX2 DevKit according to Quick Start Guide 176-0302-001 R7 (without connecting to Linux host PC) and pressed the Power button, my HTML Display showed the black screen and white texts. I did not continue with Force Recovery Mode and shut off the DevKit as already very late at night.

On next day I did the same as above (with Ethernet cable connected to live WiFi), the display showed the same black screen and white texts, but later jumped to the Ubuntu 16.04 LTS screen (without GUI menu on the left) and asked for:

nvidia
password

ubuntu
Guest Session

I typed in Password nvidia, but it showed “Invalid password”. I did the page down to another screen called Guest Session Log In, I keyed in the same password nvidia, but showed the same error.

At the same time the HDMI display flashed “System Program Problem detected”.

I presume the above error is caused by not having to carry out the Force Recovery Mode, please advise how to reset to original setting for me to complete the installation.

Appreciate please also answer to my following questions:

  1. Is Linux host PC mainly use for flashing for updates only or for something else such as programming using CUDA? Since my DevKit is connected to ethernet could it does the update. Presume after the installation it should have Ubuntu 16.04 inside the DevKit, could I just use the DevKit for development without the need of using a Linux host PC?

  2. If Linux host PC is needed, could I use Oracle Virtual Box? Alternatively I was thinking to buy a Lenovo Yoga 520-141KB laptop with touch screen Windows 10 and convert to dual boot for Windows 10 and Ubuntu (should it be version 16.04 or later version)?

  3. Where the CUDA be resident, in host PC or in DevKit? Is TX2 DevKit support CUDA 9 or later version?

  4. I am going to use the DevKit with Wireless Sensor to acquire the data, any application notes, source codes that you could advise would be most helpful including information on Wireless Sensor Network and make use of the two built-in antennas from the TX2 DevKit?

  5. Is there a way that I could include the screenshots of above issues as attachment to this reply?

Look forward to hearing from you soon for me to get going in using the TX2 DevKit.

Thank you very much for your help.

Best Regards
Hoe Min

The host (x86_64) PC is mainly for flashing the Jetson platform only. You can basically do all development work on your Jetson TX2 afterwards.

I myself have never tried to flash a Jetson platform from a VM, but other people have reported it’s workable. You can reference the following posts.

https://github.com/KleinYuan/tx2-flash
https://github.com/open-horizon/cogwerx-jetson-tx1/wiki/Initial-setup:-Jetson-TX1-and-TX2

Dear JK Chung, David, Dusty or others,

Thanks JK Chung for answering some questions, could anyone kindly answer to my following questions:

I have wired up the TX2 DevKit according to Quick Start Guide 176-0302-001 R7 (without connecting to Linux host PC) and pressed the Power button, my HTML Display showed the black screen and white texts. I did not continue with Force Recovery Mode and shut off the DevKit as already very late at night.

On next day I did the same as above (with Ethernet cable connected to live WiFi), the display showed the same black screen and white texts, but later jumped to the Ubuntu 16.04 LTS screen (without GUI menu on the left) and asked for:

nvidia
password

ubuntu
Guest Session

I typed in Password nvidia, but it showed “Invalid password”. I did the page down to another screen called Guest Session Log In, I keyed in the same password nvidia, but showed the same error.

At the same time the HDMI display flashed “System Program Problem detected”.

I presume the above error is caused by not having to carry out the Force Recovery Mode, please advise how to reset to original setting for me to complete the installation.

  1. Alternatively I was thinking to buy a Lenovo Yoga 520-141KB laptop with touch screen Windows 10 and convert to dual boot for Windows 10 and Ubuntu (should it be version 16.04 or later version)?

  2. Where the CUDA be resident, in host PC or in DevKit? Is TX2 DevKit support CUDA 9 or later version?

  3. I am going to use the DevKit with Wireless Sensor to acquire the data, any application notes, source codes that you could advise would be most helpful including information on Wireless Sensor Network and make use of the two built-in antennas from the TX2 DevKit?

  4. Is there a way that I could include the screenshots of above issues as attachment to this reply?

Look forward to hearing from you soon for me to get going in using the TX2 DevKit.

Thanks and Best Regards
Hoe Min

Are you using the password nvidia with user nvidia and not ubuntu user (which should have password ubuntu). Either nvidia/nvidia or ubuntu/ubuntu should work.

The unit ships from the factory with a limited version installed of L4T, if you are having issues, it is recommended to do a fresh install of JetPack using Force Recovery Mode.

Any machine that can install or run Ubuntu 14.04 x86_64 or Ubuntu 16.04 x86_64 should be sufficient. Others in the community have also been successful flashing JetPack from within a VM. It may not be as reliable, but if you experience failure generally it may succeed after a few attempts.

The CUDA toolkit is installed by JetPack to both the host PC and Jetson devkit. This is so you can do cross-compiling of CUDA code on the PC if you want. On the Jetson side the complete CUDA runtime and toolkit, with samples, is installed to the embedded target TX1/TX2. CUDA8 is supported by the current version of JetPack, JetPack 3.1. A future update to JetPack may provide the software upgrade to CUDA9 for Jetson TX1/TX2.

It acts as a typical WiFi/BT dual-band 2.4/5GHz wireless device, you can also have it act as an Access Point (AP mode) according to this wiki article. Ad-Hoc networking mode is not currently supported by the wireless fw according to the JetPack 3.1 release notes.

Yes, when you are composing your post, look for the Image icon on the toolbar above the text box, it’s the 2nd icon from the right and allows you to upload an image to the forum servers and inline it inside your post.

Hi Dusty_nv and others,

Happy New Year 2018.

Thanks for your explanation.

All these while my brand new TX2 DevKit flashed with error “System Program Problem Detected”.

Today managed to do sudo ./installer.sh on TX2 Dev Kit and reboot, but the same error showed on Ubuntu Desktop as well. I clicked on the Error Message Box under Report problem, it showed:

" Sorry, Ubuntu 16.04 has experienced an internal error.
If you notice further problems, try restarting the computer.
ExecutablePath
/usr/sbin/unity greater "

and I have also checked: Send an error report to help fix this problem.

Please help and advise how to solve the above-mentioned issue so I could continue to to the JetPack installer using Ubuntu 16.04 on Virtual Box 5.2.

Thanks and Best Regards
Hoe Min

Hi hoemin, the JetPack-L4T installer runs on a host PC. If using the most recent JetPack 3.2 Developer Preview, which has been released since your original post above, then Ubuntu 16.04 x86_64 is supported for the host OS. With JetPack 3.1 and prior versions, Ubuntu 14.04 x86_64 for the host OS.

Flashing from within a virtual machine isn’t officially supported, so if you are having trouble with that, try a native install for the host.
The host runs JetPack, not the Jetson. The Jetson is connected to the host running JetPack over micro-USB and Ethernet.

Dear all,

I have posted the following message, but did not receive any reply:

[i]NVIDIA Developer noreply@nvidiadeveloper.com
Sat 15/12/2018, 1:52 PM
hoemin (https://developer.nvidia.com/users/1214315) sent a message using the contact form at https://developer.nvidia.com/contact.

Hello,

I have installed Ubuntu 16.04 and JETPACK 3.3 on Linux PC and flashed to TX2 Developer Kit and can run some of the examples on TX2.

I wish to install FORTRAN compiler such as GFortran , G95 and f90 on TX2 as PGI Compilers & Tools do not support ARM processors in TX2, please advise links where I could download the above mentioned FORTRAN compiler(s) that works with TX2.

After the codes been compiled with FORTRAN compiler I would like to parallelize the FORTRAN codes on CUDA (presently installed with CUDA 9.0 in TX2), any information that you could provide will be most helpful.

Thanks.
Best Regards
HOE Min
[/i]

Since PGI Compilers and Tools do not support TX2 due to ARM processors, presume will not able to run the program written in f95 in TX2 Developer Kit, please confirm/advise.

If above method does not work, would that be possible to port the program written in C/C++ to CUDA C (part of CUDA 9.0 or need to install CUDA C separately, please advise) for parallelizing the codes using CUDA 9.0.

Appreciate your help.

Thanks and Best Regards
Hoe Min