TX2 load average too high

Autonomous Machines Jetson & Embedded Systems Jetson TX2
#1

Hello,

When I load my driver on TX2, the TX2 is getting lag . So, I typed the “top” command to check its performance. I saw the load average in per minute more than 1 (the maximum more then 2). When I insmod my driver, and the load average in per minute more than 1.3 (the maximum more than 3). Why its load average is more larger while Tx2 just reboots?
https://i.imgur.com/ecrvRJ6.png

Thanks

#2

ed_ho,

What driver are you using here? Also, could you point out which process do you think has to much load in your screenshot?

#3

Hi WayneWWW,

I am using the wireless driver in TX2. The screenshot is idle except “top” command and SSH channel.

#4

What process do you think has too much load? I don’t see any suspicious process.

#5

Hi
I am working with Ed.
Since there is no suspicious process and the system is idle, why is the “load average” more than 1 which means the system is overloading?

We are working on porting our USB3.0 802.11ad wireless to TX2.
However we have bad performance, so we check the system performance.
Before we insmod our driver, we find that the system is overloading when it is in idle

Regards
Bayer
Mr. Loop CO., LTD
office:+886-2-2648-6697 #186
Fax: +886-2-2648-6389
Cell phone:+886-912-031801
7F.-6, No.237, Sec. 1, Datong Rd., XIZHI Dist., New Taipei City 22161, Taiwan

#6

If this is what I think it is, then there isn’t really that much load. See:
https://devtalk.nvidia.com/default/topic/1032602/jetson-tx2/constant-load-average-1-0-caused-by-nvgpu_channel_p-/post/5254472/#5254472

Likely you have a process not showing up in ps, but which does participate in the load average calculation. This would be a process stuck in “disk sleep”.

FYI, disk sleep would imply a state in which the process does not show up in “ps”, but the PID does exist in “/proc//”. Such processes don’t consume any CPU resources, but seem to show up as being active (which isn’t correct).

#7