Offline cv0, cv1 and cv2

1 For CV0/1/2 offline issue
Since Jetson Orin Nano does not support CV0(DLA0 temperature), CV1(DLA1 temperature) and CV2(PVA temperature),
could them be hidden?
2. Overcurrent throttling.
Therminal#1

orinano@ubuntu:~$ sudo nvpmodel -m 2
orinano@ubuntu:~$ sudo jetson_clocks
orinano@ubuntu:~$ sudo nvfancontrol -q
FAN1:FAN_PROFILE:cool
FAN1:FAN_GOVERNOR:cont
FAN1:FAN_CONTROL:close_loop
orinano@ubuntu:~$ sudo grep “” /sys/class/hwmon/hwmon*/oc*
/sys/class/hwmon/hwmon3/oc1_event_cnt:0
/sys/class/hwmon/hwmon3/oc1_throt_en:1
/sys/class/hwmon/hwmon3/oc2_event_cnt:0
/sys/class/hwmon/hwmon3/oc2_throt_en:1
/sys/class/hwmon/hwmon3/oc3_event_cnt:0
/sys/class/hwmon/hwmon3/oc3_throt_en:1
orinano@ubuntu:~$ sudo tegrastats > tegrastats.txt
tegrastats.txt (19.4 KB)
orinano@ubuntu:~$ sudo grep “” /sys/class/hwmon/hwmon*/oc*
/sys/class/hwmon/hwmon3/oc1_event_cnt:0
/sys/class/hwmon/hwmon3/oc1_throt_en:1
/sys/class/hwmon/hwmon3/oc2_event_cnt:0
/sys/class/hwmon/hwmon3/oc2_throt_en:1
/sys/class/hwmon/hwmon3/oc3_event_cnt:807
/sys/class/hwmon/hwmon3/oc3_throt_en:1

Therminal#2

orinano@ubuntu:~$ ollama run llama3.2:3b “what problem happens to intel 14th CPUs?” | ~/piper/piper --model ~/piper/voices/english/en_US-amy-medium.onnx --output-raw | aplay -r 22050 -f S16_LE -t raw -

Hi, @energyideas

Would you mind sharing more info about the offline error?
Do you see this on the jtop? Or any error messages or logs from the system like dmesg or UI?
Could you help to share the log or screenshot with us?

We also want to know more about the working system you mentioned previously.
Could you share which device and software are using at that time?

Thanks.

Hi, @rockefellerlin

We will check with the jtop owner to see if this message can be hidden.

Thanks for sharing the data.
In your use case, the OC throttling is triggered:

/sys/class/hwmon/hwmon3/oc3_event_cnt:807

However, the GPU clocks are just slightly decreased (1019->1012).
So it’s recommended to keep using the default super mode nvpmodel config for your use case.
(you can turn off the throttling notification to let it work on the backend, please see this comment)

Thanks.

OP here. On 1/14/25 sent a private message (PM) to @AastaLLL regarding my use case and experience. No response. I follow-ed up on both Jan 20th and Jan 24 with PM, again no responses.

Hi, @energyideas

Is it possible to share the info here?
Do you meet any error or log due to the offline status?

Thanks.

Perhaps there is a language issue here, the P in PM stands for Private, not public. You still have not replied.

Based on your request, we have redirected your request.
Please wait for the corresponding team to reply.

Hi @energyideas
For clearness, please make a new post with your query. Thanks.

Got tired of waiting for this error to be fixed, purchased another one so I could finish testing out my contracted use case. Arrived yesterday , firmware dated 2023-02-21 still shipping (two years old). Did firmware update and got Jetpack6.2 on the newly arrived system.

On this new device, tried llama3.2:3b; it was working fine. Wasn’t going to try AnythingLLM because that is where I got the red ! error previously. Did try deepseek-r1. Red ! throttled. You have to reboot to get this back to MaxN. But simple inquries to deepseek-r1 still produces red ! and have to reboot.

@DaneLLL who is paid by NVIDIA suggested I start a new thread. No, I will not be starting a new thread. I am not paid by NVIDA (Mkt cap $2.91T) Fool me once, shame on you…fool me twice shame on me. Still no replies to my three PM’s.

Quotes below from @AastaLLL Please find the below two comments to turn off the notification. (either one is fine).

I tried both, one didn’t work for me, the other did. I don’t know which one. Realizing that this masks the throttling. It does eliminate the Red ! and having to reboot for increased power levels MAXN.

Off topic: Tested AI on multiple OS, computers and AI models; found it helpful to monitor cpu temperature (psensor in linux). Some setups do not regulate fan speed, and cpu temperature can immediately launch to dangerous levels.

Hope this helps.

Hi,

Thanks for the update.
If you encounter other issues, please file a new topic and we will check and follow up.