How to Calculate and Monitor Linux CPU Usage with /proc/stat, top, and pidstat

CPU Usage Calculation

Before diving into CPU usage, it is useful to understand the kernel’s jiffies counter, which increments every HZ ticks since boot. The HZ value can be read from the kernel configuration, for example:

root@node:~# grep 'CONFIG_HZ=' /boot/config-$(uname -r)
CONFIG_HZ=250

With CONFIG_HZ=250, the system records 250 ticks per second, i.e., each tick represents 4 ms. User‑space typically uses a fixed 10 ms tick.

CPU statistics are exposed in /proc/stat. A snippet of its content looks like:

root@node:~# cat /proc/stat
cpu  1549370658 9 1805063913 174156780720 9707711 0 216981980 0 725037436 0
cpu0 14624038 0 12618687 2715504522 206472 0 193511845 0 7216830 0
cpu1 12030579 0 11686870 2751300042 180759 0 10469844 0 4488307 0
cpu2 11639548 0 11156338 2752924400 140756 0 2195205 0 4473860 0
cpu3 12235160 0 11439675 2751463622 361104 0 892799 0 4916428 0

The columns represent, in order:

user (us): time spent in user mode, including guest time. nice (ni): time spent in user mode with low priority. system (sys): time spent in kernel mode. idle (id): time spent idle. iowait (wa): time waiting for I/O. irq (hi): time handling hardware interrupts. softirq (si): time handling software interrupts. steal (st): time stolen by other virtual machines (zero on bare metal). guest (guest): time spent running a virtual CPU. guest_nice (gnice): low‑priority guest time.

To compute CPU utilization, tools such as top sample the values periodically (default every 3 seconds). The usage is calculated as:

CPU% = ((Total_time_now - Idle_now) - (Total_time_prev - Idle_prev)) / (Total_time_now - Total_time_prev) * 100

This matches the algorithm used by most monitoring utilities.

Viewing CPU Usage

The most common commands are top and ps. A typical top snapshot:

top - 20:12:00 up 70 days,  3:49,  1 user,  load average: 0.10, 0.18, 0.14
Tasks: 2229 total,   1 running, 332 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.0 us,  0.0 sy,  0.0 ni, 99.9 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
KiB Mem : 52468537+total, 47777491+free, 36061900 used, 10848532 buff/cache
KiB Swap: 31250428 total, 31250428 free,       0 used. 48045312+avail Mem

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
 29973 root      20   0   42728   5972   3136 R   0.7  0.0   0:00.16 top
    9 root      20   0       0      0      0 I   0.3  0.0   0:13.52 rcu_sched
 22809 root      20   0 1024352  88780  19280 S   0.3  0.0   0:53.22 zk_proxy

The line beginning with %Cpu shows the overall CPU usage; the individual fields are explained in the previous section.

For per‑process statistics, pidstat is handy. Example:

# pidstat 1 5
Linux 4.15.0-58-generic (cs1ahyper01n07)    12/06/2025 _x86_64

10:41:09 AM   UID       PID   %usr %system %guest %CPU CPU  Command
10:41:10 AM   0        65    0.00   0.00   0.00  0.00 0    top
10:41:10 AM   0       3565   26.67  0.00   0.00 26.67 0    telegraf
10:41:10 AM   0       8788    1.90  0.00   0.00  1.90 0    mq_proxy

Field meanings: User‑mode CPU usage (%usr) Kernel‑mode CPU usage (%system) Virtual‑machine CPU usage (%guest) Total CPU usage (%CPU)

The final “Average” section of pidstat aggregates the five samples to show mean values.

This article provides the background, the raw data sources, the calculation method, and practical commands for accurately measuring CPU utilization on Linux systems.