NVIDIA drive time clock jump back

Please provide the following info (check/uncheck the boxes after creating this topic):
Software Version
DRIVE OS Linux 5.2.6
DRIVE OS Linux 5.2.6 and DriveWorks 4.0
DRIVE OS Linux 5.2.0
DRIVE OS Linux 5.2.0 and DriveWorks 3.5
NVIDIA DRIVE™ Software 10.0 (Linux)
NVIDIA DRIVE™ Software 9.0 (Linux)
other DRIVE OS version
other

Target Operating System
Linux
QNX
other

Hardware Platform
NVIDIA DRIVE™ AGX Xavier DevKit (E3550)
NVIDIA DRIVE™ AGX Pegasus DevKit (E3550)
other

SDK Manager Version
1.9.1.10844
other

Host Machine Version
native Ubuntu 18.04
other

I got this machine to set-up environment, however, it was kept unused after been purchased about 3 years.
I tried following operation

  • Connect to network and configure NTP
  • Disable NTP and manual set time date
    Problem is : time will keep jump back in to 2019-xxxx

I guess this is probably related to hardware clock battery issue.

  1. If it is the case, how should I switch the battery from the machine?
  2. Or can I just keep hardware clock disable to just use network NTP server, will it impact the time synchronization performance of the drive machine?

Please refer to Clock setting issue on Drive AGX Xavier upon set up for this.

Additionally, could you please let us know which DRIVE OS version you are using on your system?

Thanks, I will check the instruction and try to debug the issue, however, I was not following the officially instruction at first place especially on network settings which might cause some issues.

The Drive OS is 5.1.6 which I found that is no longer maintainable from official release?

There are no more releases for DRIVE AGX Xavier after DRIVE OS 5.2.6. FYI.

Understand.
From my side, I want to know if if it is possible to update from 5.1.6 up to 5.2.6 just by myside later?

I have successfully stopped the ptp server, thanks for the info.

Yes. Please see valid migration paths at Getting Started: DRIVE OS 5.2.6 :: DRIVE Platform Installation Guide with NVIDIA SDK Manager

I didn’t find the 5.1.6 in the upgrade matric, (Earliest is 5.1.12)does that means it is not supported before 5.1.12?
Btw, where can I find the release note from 5.1.6 to 5.2.6, thanks a lot!

Dear @tingguang.wang1 ,
Assuming you have DRIVE SW 10.0 on target. It has DRIVE OS 5.1.6.1. Please confirm the DRIVE OS version

Hi,

I got “5.1.6.1” via cat /etc/nvidia/version-ubuntu-rootfs.txt
When first time power on this machine, I need to install NVIDIA drive software according to OS version, which in my case should be 10, via SDK manager running on host PC, is that correct?
Is there any cmd to confirm the installed drive software version?

Dear @tingguang.wang1,
That means you already have DRIVE SW 10.0 on target. Do you want to re-flash it?

SW 10.0 is enough for now, I just want to know if it’s possible to update later.
Back to my question, how to confirm that SW 10.0 is ready? In how to set-up video, it’s necessary to install SDK manager from host PC, does that means I can skip this step and directly run perception sample code?

Yes. The DW samples must be on target. Please check /usr/local/driveworks/samples

I finally got the necessary cables to connect my target to host PC, now I am getting access to open SDKmanager via host PC(ubuntu22.04)

Based on my understanding, I need to flash my host PC side with SW10.0 to prepare the compiling environment for target board. However,it reminds me that Ubuntu22.02 is not supported by SDK manager.

Should I reset the host PC with older Ubuntu OS version?
Does pre-installed cuda, nvidia driver will give any impact to environment configuration?
Is other any alternative solution that I can directly run sample code on target borad?

Dear @tingguang.wang1,
You can run the pre-installed samples on target directly. But if you want to modify or write new DW sample to test on target, it needs to be cross compiled on host.
For cross compilation to work, you can select flashing target option in sdkmanager(you don’t need to install the SW on target, you can skip flashing the target in last step. ). It installs the required cross compilation packages on host.

You need to reset PC to old version as we don’t have docker flashing for DRIVE SW 10. Make sure you don’t have any CUDA or TRT packages on host to avoid package conflits before you setup DRIVE SW 10.0 on host.

Thanks,I found some samples under driveworks 2.2 path and it works on my target board.
However, I found the image quality is “terrify” with the default camera AR0231, probably I need to think of updating the software.

Just to confirm for cross compilation set-up, should I prepare a clean Ubuntu 20.04 or 18.04 for the host PC?

Dear @tingguang.wang1,
You need fresh Ubuntu 18.04 host.

If you want to use DW NV DNN models like drivenet, lanenet etc, They are not available in subsequent release. If you don’t need them you can upgrade to DRIVE OS 5.2.6.

Please file a new topic for issue to avoid cluttering of issues.

Thanks a lot!

Maybe just last question on this page. Sorry for that.

How do I chose btw 18.04 and 20.04. Personally, I think 20.04 could be more mature Ubuntu.
Or, it’s better just set leave a independent 18.04 for cross compiling purpose ?

Dear @tingguang.wang1,
As clarified, you need Ubuntu 18.04 host for cross compiling applications.


I prepare a host with 18.04
as picture shows, is this the expected configuration before I flash?(keep 5.1.6 at target for now)