When I want to grab multi-camera data, the vi-output occupies the cpu 100%

[jeasonli]

Hi, I’m using the samples from jetson_multimedia_api to check my cameras.
The sample which I was using is “12_camera_v4l2_cuda”, and I found an interesting phenomenon.
When I trigger /dev/video0~/dev/video3 simultaneously, the occupancy of “vi-output” on cpu raises to 100% quickly.
When I trigger /dev/video0 and /dev/video4 simultaneously, the occupancy of “vi-output” would not raise and always be approximate 7%.
The occupancy of “vi-output” will block the images update, and I can not get new images.
How can I decrease the occupancy when I trigger /dev/video0~/dev/video3, and I also want to grab images from /dev/video0 to /dev/video5 simultaneously.

[ShaneCCC]

Could you give detail reproduce steps here?

Thanks

[jeasonli]

Dear @ShaneCCC ,
Thanks for your reply.
Firstly, I comment out the code related to visualization in sample “camera_v4l2_cuda”.
And then, I open 4 terminals to run “camera_v4l2_cuda” separately. I set the “-d” from “/dev/video0” to “/dev/video3” in 4 terminals.
Next, I get high cpu usage caused by “vi-output”.
Finally, I try to use “v4l2-ctl” to check the images are dropped or not. When I use v4l2-ctl to check camera on “/dev/video0 ~ /dev/video3” simultaneously, the usage of cpu is high too, and lots of image are dropped.

[ShaneCCC]

Could you boost the clocks to try.

sudo nvpmodel -m 0
sudo jetson_clocks
sudo su
echo 1 > /sys/kernel/debug/bpmp/debug/clk/vi/mrq_rate_locked
echo 1 > /sys/kernel/debug/bpmp/debug/clk/isp/mrq_rate_locked
echo 1 > /sys/kernel/debug/bpmp/debug/clk/nvcsi/mrq_rate_locked
echo 1 > /sys/kernel/debug/bpmp/debug/clk/emc/mrq_rate_locked
cat /sys/kernel/debug/bpmp/debug/clk/vi/max_rate |tee /sys/kernel/debug/bpmp/debug/clk/vi/rate
cat /sys/kernel/debug/bpmp/debug/clk/isp/max_rate | tee  /sys/kernel/debug/bpmp/debug/clk/isp/rate
cat /sys/kernel/debug/bpmp/debug/clk/nvcsi/max_rate | tee /sys/kernel/debug/bpmp/debug/clk/nvcsi/rate
cat /sys/kernel/debug/bpmp/debug/clk/emc/max_rate | tee /sys/kernel/debug/bpmp/debug/clk/emc/rate

[jeasonli]

Dear @ShaneCCC ,
Thanks for your advise.
I have set the working mode to Max which is mode 0, and boost the clocks,
But, there are still high usage and image data dropped

[ShaneCCC]

That’s weird for the 400186200.60 fps ?!
Could you check if each sensor working well while launch individually.

Thanks

[jeasonli]

@ShaneCCC ，
Thank you for your reply.
I have checked all camera are good working individually.
The reason of 400186200.60 fps is lots of images are blocked, and they are send out in a very short period of time.
But when I active a single camera, images will not be blocked.

[ShaneCCC]

What’s the BSP version?
Did you verify on latest release J5.1.1?

[jeasonli]

@ShaneCCC ，thank you
How can I check the BSP version?
The version of jetpack which i am using is 5.0.1-b231.

[ShaneCCC]

Check by below command and the latest release is r35.3.1

cat /etc/nv_tegra_release.

[jeasonli]

Hi @ShaneCCC ,
Here is the BSP version, R35, and version details are in the image

[ShaneCCC]

I didn’t see this kind of problem with reference 6 cameras design.
From your configure didn’t looks like bandwidth cause the problem.
Suppose it could be sensor output problem cause the problem.

[jeasonli]

@ShaneCCC
Well, this issue often occurs after plugging and unplugging USB devices.
Does any other one encounter this problem?

[kayccc]

Please upgrade to the latest JetPack 5.1.1 release to see if still can reproduce this issue.