Thanks! What’s the make and model of the AC adapter you are using? Amazon Link?
Here you are: memory is 33-47% idle while playing a video on youtube (jetson nano can be seen on the bottom left of the photo):
Yes. The Amazon one that I read it was recommended for nano (Adafruit with a switch on cable to avoid plug and un plgging the jetson).
I have that one and it doesn’t work, but I had problems with it on pi as well. Possibly a bad adapter.
Can you run the top utility while decoding a video and post another screenshot?
You are right. It’s weird. I had not experienced the reboot in many days with Kodi but Youtube triggered it. It might had to do when the fact that I had three videos at a time on different tabs. They are so many variables on the table that I believe I will end up buying the barrel conector. My colleagues did not experience this issue because they use SSD disks, which less consumption (same power supply than me) but I use a hard disk of 3 Tbytes and it seems it needs more supply (and only lubuntu can stay online).
This is what you have requested me (running only a single youtube instance but other tabs still opened running in backgroupd). It has a lot of memory consumption but it does not reboot):
EDIT: Five minutes later it has rebooted. Yes: we need a barrel connector OR not connecting a hard disk with high power consumption.
From that top I can’t definiately tell if it’s using hardware decode or not. I would guess no since it’s using so much cpu. I doubt you’d be able to watch a stream at 4k smoothly, but you can try. The hardware is capable, but so far the only way I know of hardware decode on nano is with gstreamer.
Anyway, cpu or gpu decode, you managed to crash it. I’d wager a lot of things that really stress the system would do it as well. You can restrict the active cores to 2 rather than 4 cores if you use the nvpmodel command. That will probabably prevent the crashes but you lose half your cpu performance.
I will buy today the power supply you suggested on other thread. I don’t want to limit the power of jetson nano using half of the cores. Thanks so much for your comments mdegans ;)
Please help-me out as i can’t able to detect my external usb/hdd directly i need to exit from kodi and then login to ubuntu account and then re-login again with kodi account to detect usb
Please note, re Kodi, the hardware decode does not work. It’s unlikely you will be able to play Blu Ray quality video without hickups. I tested with a rip of a Blu Ray I own and it was dropping frames.
If you want hardware accelerated decode, you will have to use gstreamer for the moment.
I am curious, does nvidia have any reason for not having enabled multiarch by default for its users?
If you are speaking of both 32-bit and 64-bit user space, then this is not common for the ARM world. PCs do this because 64-bit would have had a lot of trouble being accepted without 32-bit backwards compatibility. This is the default of the Linux distributions there. On arm64 this has never been the case.
It wasn’t NVIDIA doing anything to “remove” 32-bit foreign architecture, it is just Ubuntu having never added everything required for 32-bit. In part this is very likely due to embedded systems having less storage available, and also in part because 32-bit mode on ARM has far lower performance in comparison to its 64-bit mode than what the performance hit is on a desktop PC.
You could add armhf as a foreign architecture (i386 is a foreign architecture to a 64-bit PC), add the 32-bit linker support, add 32-bit C libraries, and all 32-bit user space libraries. Then applications which are armhf should work, but I wouldn’t guarantee that there will be no difficulties or learning curve to actually do this.
Note that if you want 32-bit you could use the TK1.
the idea is to have the OS im working on for the nano (based on the latest jetpack development image) be an all in one, for not just machine learning and what nvidia intended it for, but also for a media center, and retro pc game / console game emulation unit, but when you get back into the really old games and systems they arent natively 64 bit, and some dont even have 64 bit options available, they require 32/armhf.
I’ve heard of some people attempting this with Docker, but I have never worked with this. If you can get it to work, then I suspect a Docker solution might be the better choice.