2

I test my system health by applying a cpu stress test using the stress program in a script shell. Here's the script and it's output :

xubuntu@xubuntu:~$ cat bin/test_stress.sh 

echo "=============== Starting stress test ===================="
uptime
sudo stress --cpu 4 --io 3 --timeout 2m  &
while (true)
do
sleep 5
uptime
done


xubuntu@xubuntu:~$ bin/test_stress.sh 
=============== Starting stress test ====================
 21:52:37 up 25 min,  4 users,  load average: 0.02, 0.20, 0.71
stress: info: [7384] dispatching hogs: 4 cpu, 3 io, 0 vm, 0 hdd
 21:52:42 up 25 min,  4 users,  load average: 0.34, 0.27, 0.73
 21:52:47 up 25 min,  4 users,  load average: 0.87, 0.38, 0.76
 21:52:52 up 25 min,  4 users,  load average: 1.36, 0.49, 0.80
 21:52:57 up 25 min,  4 users,  load average: 1.82, 0.60, 0.83
 21:53:02 up 25 min,  4 users,  load average: 2.23, 0.71, 0.86
 21:53:07 up 25 min,  4 users,  load average: 2.61, 0.81, 0.90
 21:53:12 up 25 min,  4 users,  load average: 2.96, 0.91, 0.93
 21:53:17 up 25 min,  4 users,  load average: 3.29, 1.01, 0.96
 21:53:22 up 25 min,  4 users,  load average: 3.59, 1.11, 1.00
 21:53:27 up 25 min,  4 users,  load average: 3.86, 1.21, 1.03
 21:53:32 up 26 min,  4 users,  load average: 4.11, 1.31, 1.06
 21:53:37 up 26 min,  4 users,  load average: 4.34, 1.40, 1.09
 21:53:42 up 26 min,  4 users,  load average: 4.56, 1.50, 1.12
 21:53:47 up 26 min,  4 users,  load average: 4.75, 1.59, 1.16
 21:53:53 up 26 min,  4 users,  load average: 4.93, 1.68, 1.19
 21:53:58 up 26 min,  4 users,  load average: 5.10, 1.77, 1.22
 21:54:03 up 26 min,  4 users,  load average: 5.25, 1.85, 1.25
 21:54:08 up 26 min,  4 users,  load average: 5.39, 1.94, 1.28

The computer heats up after running 90 secondes of the stress test and then turns off. I tried the same test on another computer for 5 minutes and it doesn't turn off. How long the computer is expected to take before it heats up and turn off during a such stress test?

Slyx
  • 213
  • 2
  • 6
  • I feel this is too broad. It depends on the type of CPU, the type of cooling methods you are using, etc. There are too many factors in play that dictate how long it _may_ take for the computer to turn off. – Tim G. Apr 24 '17 at 22:09
  • 2
    If your system is shutting itself down, during a stress test, then you have a cooling problem. That can either be caused by attempting to overclock beyond the capabilties of a cooling system or a cooling system that needs to be adjusted. A system should NEVER shut itself due to heat. – Ramhound Apr 24 '17 at 22:21
  • 1
    Its been my experience that laptops tend to have insufficient cooling capacity. High end/gaming (expensive) laptops tend to have better cooling and handle high CPU loads better. Normal laptop usage puts very little stress on the CPU. – BillThor Apr 24 '17 at 23:29
  • @BillThor. Does it mean it's normal that my laptop behaves that way? It's an HP minibook i5 vpro with 4Gb RAM. – Slyx Apr 24 '17 at 23:46
  • 1
    @Slyx Absolutely not, your laptop was designed, to run at 100% load without crashing – Ramhound Apr 25 '17 at 00:02
  • 1
    A properly configured computer will NOT shut off during a stress test. That is the goal: No shut off. The computer components should have access to sufficient cooling that they can operate safely within their stressed parameters, allowing them to initiate any performance scaling necessary to maintain their safety. Also, 5 minutes of stress isn't a real test either. Personally, when I complete a build, or when I initiate a new level of overclocking, I stress for a minimum of 8 hours, and more often a full 24 hours. I am satisfied with the build when it runs correctly the entire test. – music2myear Apr 25 '17 at 00:15
  • 2
    @BillThor some models do have issues with this, but it is a problem and a good brand will warranty their computers to NOT have this issue. 8 years ago I had a first-gen model from HP that had significant issues. It took several times shipping it to HP for repair, and basically the entire thing being replaced, before they replaced the unit as a lemon with the second-gen model. With modern components, even an insufficient cooling system should still be balanced by the self-managing capabilities of the CPU. – music2myear Apr 25 '17 at 00:17
  • @Ramhound it's a second hand computer I bought for a friend for $200. He noticed the problem just by opening two office applications on windows. I returned it back to the seller who told me he have cleaned up the cooling system and everything should work fine. In fact It work apparently fine on windows but by effectively stressing it under Linux it behaves that way. I don't know what's the correct decision to take. – Slyx Apr 25 '17 at 00:33
  • 1
    So? The laptop could cost $0 or 5 million the result is the same a laptop should not overheat after 90 seconds – Ramhound Apr 25 '17 at 01:00
  • BTW, If you know anybody selling laptops for $0 just tell me please :) – Slyx Apr 25 '17 at 01:06

1 Answers1

5

Under normal conditions, the utility stress should NOT cause a computer to overheat or power off.

Stress Utility FAQ

If the command completes successfully then you know that stress did not find any problems with your processor, I/O, memory or hard drive.

Now, you must diagnose the actual (likely physical) issue with cooling, overclocking, or power, which is outside the scope of this question.

Steven
  • 27,531
  • 11
  • 97
  • 118