htop numbers not ading up

[PsySc0rpi0n]

Hello.

I have a new issue.

I'm running Virt-Manager and there is something I'm doing that requires some I/O from an USB drive I have attached to the VM.
This stuff I'm doing requires also some CPU processing power and I was searching htop for that cpu proccessing power "taking place". I mean I was expecting to see some load on my cores but what I see is quite different.

Not only I see almost no load when I look to the top fancy and colored charts but also, the numbers shown at cpu% column on't match at all the top fancy colored charts.



At least one cpu reports 22% of load and none of the top charts reports such value!
Is there a reasonable explanation for this?

[CwF]

Yes, you only see the user. Any accurate utility would need to be root to see it all.

Just wait until you get 20GB+ memory loads and a handful of vm cores at 100% it won't make any sense at all! So basically, ignore it...

[PsySc0rpi0n]

Yes, you only see the user. Any accurate utility would need to be root to see it all.

Just wait until you get 20GB+ memory loads and a handful of vm cores at 100% it won't make any sense at all! So basically, ignore it...

No even if I run htop as root?

I didn't understand what you mean by "wait until you get 20GB+ memory loads and a handful og vm cores at 100%"

And isn't there any other tool that reports more accurate values?

[CwF]

I was thinking more of xfce's task manager that doesn't reveal real usage on my system as user dom0. The graph can show 50% cpu while no single task is above 0%. Regular 'top' ran as root right now shows one libvirt+ instance with a %cpu of 471%, and a user terminal with top does show the same. Maybe I totally misunderstood you, and jumped to my task manager thinking.
I guess my point is, I've never added things up, I just let it work!

I have a genmon line on the background of the desktop executing 'uptime' and think that is all the info I need, along with a cpu-freq plugin listing Min/Ave/Max clocks. When the load average is above 7 and min clock is over 3.5 I need to pay attention to room temperature. Up to that current limit, it's all "within the envelope" so I don't worry. with those two indicators, and the audible ramp up of fans, I usually notice when something is out of whack.

[PsySc0rpi0n]

I was thinking more of xfce's task manager that doesn't reveal real usage on my system as user dom0. The graph can show 50% cpu while no single task is above 0%. Regular 'top' ran as root right now shows one libvirt+ instance with a %cpu of 471%, and a user terminal with top does show the same. Maybe I totally misunderstood you, and jumped to my task manager thinking.
I guess my point is, I've never added things up, I just let it work!

I have a genmon line on the background of the desktop executing 'uptime' and think that is all the info I need, along with a cpu-freq plugin listing Min/Ave/Max clocks. When the load average is above 7 and min clock is over 3.5 I need to pay attention to room temperature. Up to that current limit, it's all "within the envelope" so I don't worry. with those two indicators, and the audible ramp up of fans, I usually notice when something is out of whack.

ok, I guess I just have to go with you on that. Listen for fans ramping up and not much more!

Thanks
Psy

[pendrachken]

The top bar graph is per CPU core usage, while the bottom is whole cpu% usage ( all cores together ).

The top bar graph is 39.6% when added together, meaning that you are using 39.6% of your total CPU. The bottom adds up to nearly that from what I can see of it in the thumb image - 41.7. The ~2% difference is probably just update lag from processes finishing / starting and not updating the graph more than once every $X amount of time, which is much, much longer than most time processes spend in the CPU.


To explain it better: you could have the bar graph showing 100% on core1 and 0% on the other three cores, and you would be using 25% CPU, which the bottom text part would list as "cpu%" 25.