Abnormal increase in CPU utilization

I have a relatively unusually high CPU utilization when I run applications simultaneously on the TX2 rather than on an application alone.

Please explain if this behavior can be caused by the CPU cluster migration policy.

As you can see in the example below,
The use rate for the big core is soaring,
Little core usage is not reduced.

CPU Usage, App1, App2, App1 + App2
Big (Denver) 44%, 6%, 91%
Little (A57) 49%, 38%, 88%

Best Regards,
JS Park.

What utility to get this information?

I logged using the “/proc/stats” node information just like any other cpu usage utility (htop / top …).

Best Regards,
JS Park.

It depends on what kind of apps you are using.
You can tie your app to a particular CPU.
Also, it also depends on what kind of interrupt storm the APP is raising leading to context switch.
Please check on these lines and report

Apart from above, please try these experiments also.

  1. Tie App1 to one ARM core & App2 to a Denver core.
  2. Hotplug power off all Denver cores & tie app’s to the specific ARM cores.
  3. Hotplug power off all ARM cores & tie app’s to the specific Denver cores.
    Please share results of “/proc/stats” in all cases along with original reported issue.
    Also, share more details about app & workload.

Hi truemonpark,

Any update? Is this still an issue on your end?

Thanks