How to generate CPU load on Linux server

Question

I would like to test fake load on the server, I'm looking for some burn-in or benchmark command line utility that would generate CPU load on the system.

I would like to be able to burn-in only CPU (no harddisk load, network and co) and that I would be able to set the period in which the load will run. Meaning I want something that would be able to run: CPU load for 10min on the system.

Any ideas?

ewwhite · Accepted Answer · 2013-10-16T02:24:19.300

12

You can download and install the "stress" utility... It allows you generate CPU, Memory, Disk and IO load from the command line.

Homepage: http://people.seas.harvard.edu/~apw/stress/

FAQ: http://people.seas.harvard.edu/~apw/stress/FAQ

edited Oct 16 '13 at 02:24

answered Dec 07 '10 at 10:22

ewwhite

194,921
91
434
799

score 7 · Answer 2 · answered Dec 08 '10 at 15:00

7

I prefere floating point ops:

for i in `seq 64`; do perl -e '$z=time()+(10*60); while (time()<$z) { $j++; $sqrt = sqrt($j) for (1..9999); }' & done

Be aware of your CPU number :-)

answered Dec 08 '10 at 15:00

how can I `time` this ? – Sébastien Dec 13 '12 at 11:13

PP. · Answer 3 · 2010-12-07T15:02:19.230

3

This is something I frequently want to do. But I don't have a good way of doing it. I just write a small Perl script that will just loop for as many seconds as I want.

E.g. (for 10 minutes):

perl -e '$z=time()+(10*60); while (time()<$z) { $j++; $j *= 1.1 for (1..9999); }'

Of course if you have n processors you might want to add a bash loop to create that many processes:

#!/bin/bash
# e.g. for 4 processors
for i in 1 2 3 4; do
    perl -e .... &
done
wait

edited Dec 07 '10 at 15:02

answered Dec 07 '10 at 08:55

PP.

3,246
6
26
31

4

Minor point: Won't that `for` loop run them in series, rather than parallel? I think `(perl -e '...' &)` to detach them from the shell would do the job, though. – SmallClanger Dec 07 '10 at 09:36
You're right. I'll correct my post. Thanks for pointing that out. – PP. Dec 07 '10 at 15:02

score 1 · Answer 4 · answered Dec 07 '10 at 08:51

1

cpuburn

"How do I run a command, and have it abort (timeout) after N seconds?"

answered Dec 07 '10 at 08:51

Ignacio Vazquez-Abrams

45,019
5
78
84

score 1 · Answer 5 · answered Nov 26 '15 at 20:12

1

I recommend using stress-ng, it has a lot more features than stress and can exercise far more of the operating system.

http://kernel.ubuntu.com/~cking/stress-ng/

answered Nov 26 '15 at 20:12

Colin King

21
1

score 1 · Answer 6 · edited Mar 20 '17 at 10:16

Here is a pure shell (at least bash & ksh) way that I already posted as a reply to a similar question here

# Usage: lc [number_of_cpus_to_load [number_of_seconds] ]
lc() {
  (
    pids=""
    cpus=${1:-1}
    seconds=${2:-60}
    echo loading $cpus CPUs for $seconds seconds
    trap 'for p in $pids; do kill $p; done' 0
    for ((i=0;i<cpus;i++)); do while : ; do : ; done & pids="$pids $!"; done
    sleep $seconds
  )
}

How to generate CPU load on Linux server

6 Answers6

Linked