R7800 CPU frequencies

[LeKeiser]

Hello everyone,

Stupid question...
When I check my R7800 status with Kamoj Addon, I get this :

CPU Frequencies	Core 0: 1.725 GHz (100.0 %) Core 1: 1.725 GHz (100.0 %) (384MHz 600MHz 800MHz 1.000GHz 1.400GHz 1.725GHz)
	CPU Governors	Core 0: ondemand 79% Core 1: ondemand 10% (interactive conservative ondemand userspace powersave performance)
	CPU Load Total (processes) (per core)	2.6%, p:45 ( 3.2% p:21, 2.0% p:24 ) Threads: 143 (System Load Average last: 1/5/15 minutes: 4.14 / 3.18 / 1.54)

The first line, does it mean that the 2 cores are running full steam, 1.725GHz, or that it is the highest frequency they can run ?
Or is the first core running at 79% and the second one 10% ?
I tried to check the TOP information :
%Cpu0 _: _ 0.0/5.3 _ _ 5[||||| _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ]
%Cpu1 _: _10.5/10.5 _ 21[|||||||||||||||||||||| _ _ _

Isn't the load quite high ? Like very high ?
top-procps-ng - 21:17:39 up _1:13, _0 users, _load average: 4.18, 4.16, 4.14

_PID USER _ _ _PR _NI _ _VIRT _ _RES _%CPU _%MEM _ _ TIME+ S COMMAND
19869 root _ _ _20 _ 0 _ _1.6m _ 0.6m _11.8 _ 0.1 _ 0:00.03 S /bin/sh /sbin/wlan stainfo
19936 root _ _ _20 _ 0 _ _1.9m _ 0.8m _11.8 _ 0.2 _ 0:00.04 R /usr/bin/top-procps-ng -b -n1 -c -w512
_ 32 root _ _ _20 _ 0 _ _0.0m _ 0.0m _ 5.9 _ 0.0 _ 2:25.56 S [kworker/0:1]
_ _3 root _ _ _20 _ 0 _ _0.0m _ 0.0m _ 0.0 _ 0.0 _ 0:01.83 S [ksoftirqd/0]

CPU doesn't seem overloaded to me...

Not sure what to make of it...
CPU is running at 46°C, a bit warmer than earlier this week, it was around 42° (I have a fan below the router)

So, what do you think ?
Ah, running latest @Voxel l firmware and latest @kamoj Addon

Thanx for your feedback.

Cheers.

 

[kamoj]

First line means that the CPU is clocked with highest frequency.

The governor tells that core 0 is used to 79% and core 1 to 10%.
(The governor is used to automatically clock down cpu's at low load to keep down power usage and temperature.)

The cpu load in the R7800 is about 3.88 too high when it shows >=3.9.
It's an old bug that has always been there in the original Netgear Firmware.

 

[LeKeiser]

Hello Mr @kamoj

So the first line means that the CPU is running full steam, is that correct ?
At the moment, TOP gives me this :
top-procps-ng - 17:06:36 up 15:28, _0 users, _load average: 4.05, 4.18, 4.17
If I'm not mistaken, the CPUs are running 200% and more, right ? But the router is doing...nothing. I mean, not that I know of.
_PID USER _ _ _PR _NI _ _VIRT _ _RES _%CPU _%MEM _ _ TIME+ S COMMAND
8020 root _ _ _20 _ 0 _ _0.0m _ 0.0m _ 5.9 _ 0.0 _12:20.20 S [kworker/0:2]
5277 root _ _ _20 _ 0 _ _1.9m _ 0.8m _ 5.9 _ 0.2 _ 0:00.09 R /usr/bin/top-procps-ng -b -n1 -c -w512
_ _2 root _ _ _-2 _ 0 _ _0.0m _ 0.0m _ 0.0 _ 0.0 _ 0:00.03 S [kthreadd]

The PS command shows me nothing.

I'm a bit worried. I would like to know what is forcing the CPUs to run so high. Something in the network ?

Or are you saying that there I a bug in the original Netgear firmware and I shouldn't worry at all ?

Thanx for your feedback

 

[kamoj]

The cpu load in the R7800 is about 3.88 too high when it shows >=3.9.
(So to get correct value, please subtract 3.88 from all values >3.9).
You can also try the command:

 mpstat -P ALL 1 1

It's an old bug that has always been there in the original Netgear Firmware.

 

[LeKeiser]

Again, thanx a lot for your answer Mr @kamoj
I guess everything is normal, no need to worry. Right ?

root@R7800:/$ mpstat -P ALL 1 1
Linux 3.4.103 (R7800) 07/10/23 _armv7l_ (2 CPU)

06:42:34 CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle
06:42:35 all 13.37 0.00 11.88 0.00 0.99 1.98 0.00 0.00 0.00 71.78
06:42:35 0 21.00 0.00 8.00 0.00 1.00 3.00 0.00 0.00 0.00 67.00
06:42:35 1 5.88 0.00 15.69 0.00 0.98 0.98 0.00 0.00 0.00 76.47

Average: CPU %usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle
Average: all 13.37 0.00 11.88 0.00 0.99 1.98 0.00 0.00 0.00 71.78
Average: 0 21.00 0.00 8.00 0.00 1.00 3.00 0.00 0.00 0.00 67.00
Average: 1 5.88 0.00 15.69 0.00 0.98 0.98 0.00 0.00 0.00 76.47

 

[kamoj]

All is fine.
Do you have heavy network traffic to worry about?

 

[LeKeiser]

All is fine.
Do you have heavy network traffic to worry about?

Nahh it's a home router with only me in the house. Around 25-30 devices max (Echos, Google Home, wifi outlets, iPads, AppleTV, Shield, ...). Pretty normal house
Thanx for your answers, I'm reassured now

 

[sfx2000]

Isn't the load quite high ? Like very high ?
top-procps-ng - 21:17:39 up _1:13, _0 users, _load average: 4.18, 4.16, 4.14

One cannot apply the same metrics to an embedded linux that one does for desktop linux...

Also should note that busybox "top" handles things different than procps, so you'll get different results... busybox looks at total CPU, whereas procps and gnu top look at cores/processes/threads to calculate values..

Also consider that top (or busybox top) samples over time - need to keep sampling rates low, otherwise top would dominate the task list...

It's more important for relative comparisons on the same HW/SW, not across platform, esp in embedded space...