Temperature above threshold, cpu clock throttled

deathsfriend99 · 10-24-2008, 03:43 PM

I am not sure if this is a hardware of software issue, but here it goes. I installed CentOS 5 on a machine that was running FC4. It now lockes up and gives the log message:
kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 686)
kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 686)

I found it odd that there was no overheating issues before I went to CentOS 5 so I checked lm_sensors.

# sensors
No sensors found!

So I wonder where acpi is getting this temperature info from.

It appears to be unsupported, so why is it complaining?
# cat /proc/acpi/processor/CPU0/limit
<not supported>
# cat /proc/acpi/processor/CPU0/throttling
<not supported>

It is a Dell with no bios info on temp, so I can't see if it's really overheating, but it shouldn't be attempting to throttle if it doesn't support throttling

Anyone heard of this?

irishbitte · 10-24-2008, 04:28 PM

try booting and passing the kernel option NOACPI at boot? I have heard of this before, but it is relatively rare! sounds like something is broken in there. BTW, NOACPI means that the fan will probably run all the time, but it will take away the cpu throttling symptom. I reckon a hardware problem, maybe....

x_terminat_or_3 · 10-24-2008, 07:06 PM

I have seen this before, I also have a Dell computer.

When I forgot to hook up the cpu fan after I moved the motherboard to a different case, I saw those messages as well. Even if lm_sensors can't detect any sensors, it doesn't mean that the CPU doesn't know what temperature it is running at. The cpu fan is controlled almost directly by the cpu, the hotter it gets, the harder the fan spins. On Dell boxes, the cpu fans are quite powerfull and usually noisy if running hot.

I suggest you verify that the cpu fan is running correctly before trying anything else. To do this, boot up your pc check the back of your pc. There should be wind blowing out of the box where your power cable goes in (your PSU), but there should also be wind blowing from a blow hole a bit below that.

Or if you're feeling adventurous, open your case, usually easy with Dells, turn on your pc and visually see that the cpu fan is running.

deathsfriend99 · 10-24-2008, 08:17 PM

I have checked the cpu fan, and it is working great. The heatsink appears to be pretty clean also. I disabled acpid at startup. I assumed that's where the error messages originate from. Hopefully that will solve the problem.

jiml8 · 10-24-2008, 08:24 PM

You really need to find out what the temperature actually is. Turning off the warnings is akin to disconnecting the oil pressure light in your car. Doesn't stop the problem, just prevents you from knowing about it.

deathsfriend99 · 10-24-2008, 08:30 PM

I agree, but I don't know how to do that. There is no sensors, the bios has no temp info. This old dell has a pretty useless bios. It would be strange that it just started giving over temp warnings when acpi has no temp info in it, and it just started the day after I installed Centos5.

deathsfriend99 · 10-28-2008, 02:31 PM

I booted with acpi=off and I still get the error. I am even more convinced that this is an anomalous reading. The fan and heatsink are clean and running fine. I can't seem to find anywhere where acpi is getting a temp from. Sensors doesn't work, gkrellm shows no cpu temp, cat /proc/acpi/processor/CPU0/throttling
<not supported>

I don't know where to move on from here. This workstation is useless since it locks up every hour or so do to throttling that I can't seem to disable.

deathsfriend99 · 10-29-2008, 01:22 PM

At this point, I'm pretty sure I'm talking to myself, but hopefully this will help someone in the future. I think I have found the problem. This machine was loaded by someone else, and it seems a laptop module was loading (asus_acpi). Doing an lsmod showed it. I had no use for anything asus since this is a crappy dell desktop and has zero chance of having an asus mb much less needing notebook modules.

After removing it and blacklisting it, I have not seen the problem return.

irishbitte · 10-29-2008, 03:25 PM

Very interesting! Unusual that a laptop module was loaded! Sorry for not posting for a bit, was away. Well done on finding the issue.

kromberg · 10-29-2008, 04:11 PM

Most software on RHEL5 is about 2 to 3 years out of date, but it is "secure". The version of lm_sensors is way out of date.

sunilshah78 · 12-08-2008, 03:42 AM

I am not sure if this is a hardware of software issue, but here it goes. I installed CentOS 5 on a machine that was running FC4. It now lockes up and gives the log message:
kernel: CPU0: Temperature above threshold, cpu clock throttled (total events = 686)
kernel: CPU1: Temperature above threshold, cpu clock throttled (total events = 686)

I found it odd that there was no overheating issues before I went to CentOS 5 so I checked lm_sensors.

# sensors
No sensors found!

So I wonder where acpi is getting this temperature info from.

It appears to be unsupported, so why is it complaining?
# cat /proc/acpi/processor/CPU0/limit
<not supported>
# cat /proc/acpi/processor/CPU0/throttling
<not supported>

It is a Dell with no bios info on temp, so I can't see if it's really overheating, but it shouldn't be attempting to throttle if it doesn't support throttling

Anyone heard of this?[/QUOTE]

x_terminat_or_3 · 12-08-2008, 01:37 PM

As I mentioned before, your CPU knows its temperature.

If the temperature is too high, your CPU will throw an exception. Linux has code to handle this, and takes the appropriate action (throttling down = less power = less temperature)

sunilshah78, Welcome to LinuxQuestions, please read the rules, and, also, note, that it is generally a bad idea to high-jack other people's questions.