I’ve have followed everything and installed from SDK half a dozen times carefully checking I have the latest versions. All runs well but after 30 mins or so the above two processes crash. I have seen this reported elsewhere but don’t know how to diagnose and fix. I have carefully reformatted both an SD and NVME and tried both with multiple fresh installs with new SDK downloads. Any ideas?
Just a wild guess, try doing a native build and see if that clears it up. I have never used those packages. However, it seems like those are targeting amd64 platforms, some of this stuff does play differently and that might be part of it.
1 Like
Many thanks. I have absolutely no idea how to do that but will get researching!
1 Like
That might be a waste of time.
I just now went on the desktop and it generated a ubuntu crash report for tracker…
Did not see it before because I am running then dev board remotely.
1 Like
Well I guess it’s not just me then! But whose bug is it? NVIDIA?
You have look at the entire ecosystem.
End game is to sell you subscription services and “containers”, hence the current state of support for the product.
So far I managed to find the micro small details that are essential to having success with dev on this board.
Notice how they dropped the street price to $249 dollars, we just paid almost $600 for what?? Spending so much time trying to do the basics.
The money is in the container and enterprise level support. If they told you how to fix the problem on this forum they would not sell any support subscriptions…
As far as the tracker-3, that is a ubuntu issue, we have seen that before on desktops. I appears to be benign, I was testing the board and working on it remotely and did not run into any brick walls that were attributed to it.
fred@orin1:~/save-downloads$ systemctl --user list-units | grep tracker
tracker-miner-fs-3.service loaded active running Tracker file system data miner
fred@orin1:~/save-downloads$ systemctl --user status tracker-extract-3.service
○ tracker-extract-3.service - Tracker metadata extractor
Loaded: loaded (/usr/lib/systemd/user/tracker-extract-3.service; static)
Active: inactive (dead)
Dec 31 21:33:23 orin1 systemd[1178]: Starting Tracker metadata extractor...
Dec 31 21:33:24 orin1 systemd[1178]: Started Tracker metadata extractor.
Dec 31 22:12:58 orin1 systemd[1178]: Starting Tracker metadata extractor...
Dec 31 22:12:59 orin1 systemd[1178]: Started Tracker metadata extractor.
Dec 31 22:32:54 orin1 systemd[1178]: Starting Tracker metadata extractor...
Dec 31 22:32:55 orin1 systemd[1178]: Started Tracker metadata extractor.
fred@orin1:~/save-downloads$ systemctl --user stop tracker-extract-3.service
Ok thanks. I’m afraid I’m one of the $249 guys! Well actually £199 UK + VAT = £249! I’m not developing a product but just wanted to see if I could put a smart ip security camera system together with object recognition/detection. I saw Jensen Huang’s vid last week and think I snagged the last Orin Nano in the UK. Maybe I’ll eBay it and get a Pi 5 with IR board! I would rather stick with it and get it going though. Happy New Year BTW. 😂
Pi5, is not even close to same class as this board.
The value is in the GPU, we are using edge detection for a parts loader I designed and this is the only little board that can handle it. I can use a big box Intel board with a GPU but this makes for a better fit in the control cabinet and much easier to mount and simple power requirements.
We are so far behind on this now due to the snags its not even funny. I can say with 100% confidence that if this board did not have a GPU it would have been in the trash on day 2.
1 Like
I did make a typo. I meant to put AI board not IR board as there is now a Hailo board extension for Pi5. 26 tops if that is enough? Raspberry Pi AI HAT+ | The Pi Hut
The pi5 we have here had the cam on csi, it was very slow when your are in the lens and can see the monitor. If you are using it remotely it would not be noticed. This nano has a USB brio 510 connected and a csi cam running at the same time, very good performance on even the USB. Running a Brio 510 and it is so much better than the imx219 cam and 4xx something cam. I am also using the nv libs that are optimized. If you have the time and patience to get the nano up you will be have a better platform to work from. My attention span is comparable to a 2 year child, its not gotten much better after all these years. It just depends on your willingness to keep working on it.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.