Tegrastats missing information when run inside a container

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
1

I’m trying to run tegrastats inside a container, but some info are missing. Here is what I obtain when running on host:


10-30-2024 17:57:06 RAM 10526/15656MB (lfb 2x4MB) CPU [43%@1984,43%@1984,40%@1984,44%@1984,38%@1344,47%@1344,41%@1344,44%@1344] GR3D_FREQ 80% cv0@64.781C cpu@66.968C soc2@65.343C soc0@65.718C cv1@66.375C gpu@64.375C tj@67.093C soc1@67.093C cv2@64.968C VDD_IN 11826mW/11015mW VDD_CPU_GPU_CV 5440mW/4716mW VDD_SOC 2135mW/2128mW

And inside the container:


10-30-2024 17:57:06 RAM 10526/15656MB (lfb 2x4MB) CPU [40%@1574,42%@1574,42%@1574,44%@1574,45%@1984,48%@1984,39%@1984,42%@1984] EMC_FREQ 0%@2133 GR3D_FREQ 0%@[0] NVENC1 off NVDEC1 off APE 174 cv0@65C cpu@67.187C soc2@65.375C soc0@65.531C cv1@66.562C gpu@64.25C tj@67.187C soc1@66.937C cv2@64.937C VDD_IN 11984mW/11766mW VDD_CPU_GPU_CV 5598mW/5420mW VDD_SOC 2135mW/2135mW

I’m running the container as privileged and here are the volumes that I mount inside:

/usr/bin/jetson_clocks
/sys
/proc/meminfo
/proc/stat
/etc/localtime
/etc/nv_tegra_release

It just seems that there are missing volumes to mount, but can’t figure out which.

Thanks

4

Hi,

Could you help to point out which info is missing inside the container?
The logs that you shared with us seem to capture the same info.

Thanks.

5

My bad. I updated the output.

The EMC_FREQ & GR3D_FREQ are always 0 inside the container.

6

Hi,

GPU loading info is captured from the below node:

sys/devices/platform/bus@0/17000000.gpu/load

Could you try to mount the node to see if you can get the expected tegrastats output inside the container?

Thanks.

7

@AastaLLL I tried mounting the volume
I can see it inside, but tegrastats is still returning 0 values when inside the container

8

If you mount the /bin/ folder as a volume, you should be able to run tegrastats:

docker run -it --rm -v /bin:/bin your-image-name
9

@clarkca1 I’m actually able to run tegrastats, but data related to GR3D and EMC are set to 0, and do not correspond to the real values ( returned by tegrastats run on the host).

10

Hi,

We need to check it further to see what info is required to show GR3D_FREQ.
Will provide more info to you later.

Thanks.

12

Hi,

Sorry for the late update.
GPU status is available once the whole /sys node is mounted.

$ sudo docker run -it --rm --net=host --runtime nvidia -v /bin:/bin -v /sys:/sys -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/l4t-cuda:12.6.11-runtime

root@tegra-ubuntu:/# tegrastats 
...
12-19-2024 04:00:55 RAM 13420/62841MB (lfb 27x4MB) SWAP 0/31420MB (cached 0MB) CPU [0%@2201,100%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201] EMC_FREQ 14%@3199 GR3D_FREQ 99%@[1276,1276] NVENC off NVDEC off NVJPG off NVJPG1 off VIC off OFA off NVDLA0 off NVDLA1 off PVA0_FREQ off APE 174 cpu@55.718C soc2@51.843C soc0@51.937C gpu@52.25C tj@55.718C soc1@50.25C VDD_GPU_SOC 18792mW/5552mW VDD_CPU_CV 2801mW/1627mW VIN_SYS_5V0 6360mW/3921mW
12-19-2024 04:00:56 RAM 13420/62841MB (lfb 27x4MB) SWAP 0/31420MB (cached 0MB) CPU [0%@2201,100%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201] EMC_FREQ 15%@3199 GR3D_FREQ 99%@[1275,1275] NVENC off NVDEC off NVJPG off NVJPG1 off VIC off OFA off NVDLA0 off NVDLA1 off PVA0_FREQ off APE 174 cpu@55.75C soc2@52.218C soc0@52.156C gpu@52.406C tj@55.75C soc1@50.343C VDD_GPU_SOC 18792mW/5680mW VDD_CPU_CV 2800mW/1639mW VIN_SYS_5V0 6360mW/3945mW
...

Thanks.

14

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.