Load Average on Luckfox Pico Ultra W RV1106 with Buildroot

2025-06-01 19:00

Hello,

I am working with a Luckfox Pico Ultra W board (RV1106) running a Buildroot system.
I've noticed a consistently very high load average (e.g., Load average: 10.24 10.10 10.11) when I check using the top command, even when no specific user applications are running (i.e., the system is mostly idle). At the same time, top shows a high CPU idle percentage (around 85-95%), with low user and system CPU usage.

Code: Select all

Mem: 161076K used, 14140K free, 880K shrd, 7808K buff, 127000K cached
CPU:   1% usr   3% sys   0% nic  95% idle   0% io   0% irq   0% sirq
Load average: 10.04 10.06 10.08 1/97 343
  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
 3755  2529 root     R     1800   1%   0% top
 2519   311 root     S     5276   3%   0% sshd: root@pts/1
  431     2 root     SW       0   0%   0% [irq/67-dwc3]
  272     2 root     DW       0   0%   0% [vmcu]
  574     1 root     S    33016  19%   0% smbd -D
  577   574 root     S    32088  18%   0% {smbd-notifyd} smbd -D
  578   574 root     S    32080  18%   0% {smbd-cleanupd} smbd -D
  629     1 root     S    24604  14%   0% nmbd -D
  415     1 root     S    10220   6%   0% /usr/bin/adbd
  460     1 root     S     5192   3%   0% rkwifi_server start
  311     1 root     S     4964   3%   0% sshd: /usr/sbin/sshd [listener] 0 of 10-100 startups
  183     1 root     S     4928   3%   0% /usr/libexec/bluetooth/bluetoothd -n
 2529  2519 root     S     1828   1%   0% -sh
  177     1 dbus     S     1812   1%   0% dbus-daemon --system
 1333     1 root     S     1804   1%   0% {S99usb0config} /bin/sh /etc/init.d/S99usb0config start
    1     0 root     S     1796   1%   0% init
 1334     1 root     S     1792   1%   0% /sbin/getty -L console 0 vt100
   86     1 root     S     1788   1%   0% /sbin/syslogd -n
  325     1 root     S     1788   1%   0% /usr/sbin/telnetd -F
   90     1 root     S     1784   1%   0% /sbin/klogd -n
  101     1 root     S     1424   1%   0% /sbin/udevd -d
  565     1 root     S     1248   1%   0% wpa_supplicant -B -i wlan0 -c /data/wpa_supplicant.conf -d
  643     1 root     S      956   1%   0% hciattach -s 1500000 /dev/ttyS1 any 1500000 flow nosleep
 1314     1 root     S      768   0%   0% luckfox_switch_rgb_resolution
  254     2 root     SW       0   0%   0% [vcodec_thread_0]
  250     2 root     SW       0   0%   0% [queue_work0]
    7     2 root     SW       0   0%   0% [ksoftirqd/0]
 3341     2 root     IW<      0   0%   0% [kworker/0:0H-kb]
   11     2 root     SW       0   0%   0% [kcompactd0]
   48     2 root     SW       0   0%   0% [kswapd0]
  268     2 root     DW       0   0%   0% [rkisp-vir0]
 4065     2 root     IW       0   0%   0% [kworker/0:1-eve]
   40     2 root     SW       0   0%   0% [kconsole]
  425     2 root     SW       0   0%   0% [aicwf_busrx_thr]
  267     2 root     DW       0   0%   0% [venc]
  423     2 root     SW       0   0%   0% [aicwf_pwrctl]
  252     2 root     SW       0   0%   0% [queue_work1]
  468     2 root     IW       0   0%   0% [kworker/u2:2-ev]
  253     2 root     SW       0   0%   0% [irq/50-ffa60000]
   67     2 root     SW       0   0%   0% [jbd2/mmcblk0p7-]
  151     2 root     SW       0   0%   0% [jbd2/mmcblk0p5-]
  117     2 root     IW<      0   0%   0% [kworker/0:3H-kb]
   56     2 root     IW<      0   0%   0% [stmmac_wq]
  175     2 root     SW       0   0%   0% [irq/51-ffa70000]
   45     2 root     IW<      0   0%   0% [kworker/u3:0-hc]

This high idle load average seems unusual, and I'm concerned it might be an underlying issue affecting the performance of applications I intend to run, such as real-time video streaming with WebRTC, where I've been experiencing significant lag and freezes.

My questions are:

1. Is such a high load average (around 10) with high CPU idle considered normal or expected for this board with a standard Buildroot configuration?
2. If this is not normal, what could be the common causes for this behavior on the RV1106 platform (e.g., I/O wait issues, specific kernel tasks, driver problems, default services)?
3. Are there any recommended diagnostic tools or steps specific to the Luckfox Pico Ultra W / RV1106 to investigate this high idle load average further?
Any insights or suggestions would be greatly appreciated!

Thanks,

2025-06-03 7:32

Hello, the performance optimization of the system is not within the scope of technical support. For your issue, I can only offer some simple suggestions. For more details, please conduct your own tests and confirm.
1. The load average can be used to represent (not equivalent) the length of the process queue waiting for scheduling . These processes may just be waiting for a certain resource or lock. If they do not excessively occupy CPU and memory, the system can operate normally. In actual tests, when the default image of load average is around 10, RV1106 can still run the rkips routines that require more CPU scheduling. This can be used to determine that there is no direct connection between load average and system lag.
2. The high value of load average is mainly affected by the rockit driver. If you need the load average to be at a lower level, you cannot use the operations related to hardware coding. The source code of rockchip's driver is not open source, so it is impossible to further determine the reason for the high load average value.
3. We have no relevant experience for this issue and cannot provide suggestions.

2025-06-08 15:39

Thank you very much.