This is linked to this question.

When I run top I get the following result:

enter image description here

pid 3038 is using 18% cpu, however when running

enter image description here

the result is 5.5%. And this number does not appear to be changing with time (i.e. when running the same command a bit later)...

Is the ps command somehow averaging the cpu usage?


2 Answers 2


man ps in NOTES section.

   CPU usage is currently expressed as the percentage of time spent running
   during the entire lifetime of a process.  This is not ideal, and it does not
   conform to the standards that ps otherwise conforms to.  CPU usage is
   unlikely to add up to exactly 100%.

And, guess you know, but you can also do:

top -p <PID>

Edit: as to your comment on other answer;

"Hmm yeah i Wonder how to get that (the instant CPU percentage) from ps"

Short answer: you can't.

Why is it so?

It is like asking someone to calculate the speed of a car from a picture.

While top is a monitoring tool, ps is a snapshot tool. Think of it like this: At any given moment a process either uses the CPU or not. Thus you have either 0% or 100% load in that exact moment.

Giving: If ps should give instant CPU usage it would be either 0% or 100%.

top on the other hand keep polling numbers and calculate load over time.

ps could have given current usage – but that would require it to read data multiple times and sleep between each read. It doesn't.

Calculation for ps %cpu

ps calculates CPU usage in the following manner:

uptime  = total time system has been running.
ps_time = process start time measured in seconds from boot.
pu_time = total time process has been using the CPU.

;; Seconds process has been running:
seconds   = uptime - ps_time
;; Usage:
cpu_usage = pu_time * 1000 / seconds

print: cpu_usage / 10 "." cpu_usage % 10

Example: uptime = 344,545 ps_time = 322,462 pu_time = 3,383 seconds = 344,545 - 322,462 = 22,083 cpu_usage = 3,383 * 1,000 / 22,083 = 153 print: 153 / 10 "." 153 % 10 => 15.3

So the number printed is: time the process has been using the CPU during it's lifetime. As in the example above. It has done so in 15.3% of its lifetime. In 84,7% of the time it has not been bugging on the CPU.

Data retrieval

ps, as well as top, uses data from files stored under /proc/ - or the process information pseudo-file system.

You have some files in root of /proc/ that have various information about the overall state of the system. In addition each process has its own sub folder /proc/<PID>/ where process specific data is stored. So, for example the process from your question had a folder at /proc/3038/.

When ps calculates CPU usage it uses two files:

/proc/uptime      The uptime of the system (seconds), and the amount of time spent in idle process (seconds).
/proc/[PID]/stat  Status information about the process.
  • From uptime it uses the first value (uptime).
  • From [PID]/stat it uses the following:
 #  Name      Description
14  utime     CPU time spent in user code, measured in jiffies
15  stime     CPU time spent in kernel code, measured in jiffies
16  cutime    CPU time spent in user code, including time from children
17  cstime    CPU time spent in kernel code, including time from children 
22  starttime Time when the process started, measured in jiffies

A jiffie is clock tick. So in addition it uses various methods, ie., sysconf(_SC_CLK_TCK), to get system's Hertz (number of ticks per second) - ultimately using 100 as a fall-back after exhausting other options.

So if utime is 1234 and Hertz is 100 then:

seconds = utime / Hertz = 1234 / 100 = 12.34

The actual calculation is done by:

total_time = utime + stime

IF include_dead_children
    total_time = total_time + cutime + cstime

seconds = uptime - starttime / Hertz

pcpu = (total_time * 1000 / Hertz) / seconds

print: "%CPU" pcpu / 10 "." pcpu % 10

Example (Output from a custom Bash script):

$ ./psw2 30894
System information
           uptime : 353,512 seconds
             idle : 0
Process information
              PID : 30894
         filename : plugin-containe
            utime : 421,951 jiffies 4,219 seconds
            stime : 63,334 jiffies 633 seconds
           cutime : 0 jiffies 0 seconds
           cstime : 1 jiffies 0 seconds
        starttime : 32,246,240 jiffies 322,462 seconds

Process run time  : 31,050
Process CPU time  : 485,286 jiffies 4,852 seconds
CPU usage since birth: 15.6%

Calculating "current" load with ps

This is a (bit?) shady endeavour but OK. Lets have a go.

One could use times provided by ps and calculate CPU usage from this. When thinking about it it could actually be rather useful, with some limitations.

This could be useful to calculate CPU usage over a longer period. I.e. say you want to monitor the average CPU load of plugin-container in Firefox while doing some Firefox-related task.

By using output from:

$ ps -p -o cputime,etimes

cputime TIME     cumulative CPU time, "[DD-]hh:mm:ss" format.  (alias time).
etime   ELAPSED  elapsed time since the process was started, [DD-]hh:]mm:ss.
etimes  ELAPSED  elapsed time since the process was started, in seconds.

I use etime over etimes in this sample, on calculations, only to be a bit more clear. Also I add %cpu for "fun". In i.e. a bash script one would obviously use etimes - or better read from /proc/<PID>/ etc.

$ ps -p 30894 -o %cpu,cputime,etime,etimes
 5.9 00:13:55    03:53:56   14036

 6.2 00:14:45    03:56:07   14167

Calculate times:
            13 * 60 + 55 =    835   (cputime this far)
3 * 3,600 + 53 * 60 + 56 = 14,036   (time running this far)

            14 * 60 + 45 =    885   (cputime at end)
3 * 3,600 + 56 * 60 +  7 = 14,167   (time running at end)

Calculate percent load:
((885 - 835) / (14,167 - 14,036)) * 100 = 38

Process was using the CPU 38% of the time during this period.

Look at the code

If you want to know how ps does it, and know a little C, do (looks like you run Gnome Debain deriavnt) - nice attitude in the code regarding comments etc.:

apt-get source procps
cd procps*/ps
  • I might update with more information on top, and continuous monitoring - or snap by delay aka "ps" with current CPU load.
    – Runium
    Dec 16, 2012 at 10:45
  • 5
    Really great answer!
    – Theodor
    Dec 16, 2012 at 11:11
  • What about when calculating total CPU usage across the entire system. Are tools trying to get the CPU usage via a snapshot, or averaging it over time? Jul 19, 2014 at 8:20
  • How does the top -bn1 command calculates the %CPU column?
    – FedKad
    Jun 11, 2021 at 16:55
man top

%CPU  --  CPU usage
The  task's share of the elapsed CPU time since the last screen update, expressed as a percentage of total CPU time.  In a true SMP environment, if 'Irix
mode' is Off, top will operate in 'Solaris mode' where a task's cpu usage will be divided by the total number of CPUs.  You toggle  'Irix/Solaris'  modes
with the 'I' interactive command.

man ps 
%cpu       %CPU    cpu utilization of the process in "##.#" format. Currently, it is the CPU time used divided by the time the process has been running                       (cputime/realtime ratio), expressed as a percentage. It will not add up to 100% unless you are lucky. (alias pcpu).
  • Hmm yeah i Wonder how to get that (the instant CPU percentage) from ps
    – Theodor
    Dec 15, 2012 at 19:11
  • 3
    you can extract from top example top -p 3343 -n1 | awk '/ R /{print $10}' Dec 15, 2012 at 19:31
  • 1
    Awesome. I found that awk:ing for the pid worked better, as in top -p 3343 -n1 | awk '/ 3343 /{print $10}'
    – Theodor
    Dec 16, 2012 at 11:06
  • 2
    Thanks "top -p" is nice, I never noticed this option! However I dont like searching for the PID, what if by chance the same digits appear in the free memory section or something. On my system the information is on the 8th line and 9th column hence I do: top -p $PID -n1 | awk '{if (NR ==8) print $9 }'
    – phil_w
    Apr 13, 2015 at 16:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .