Package temperature above threshold, cpu clock throttled

Question

I am running 64 bit Ubuntu 11.10 on an i7 with 8gigs of ram. I thought of putting this on askubuntu.com but decided that maybe the question has a much broader appeal.

I have the following error message popping up when I run math simulations.

CPUn: Core temperature above threshold, cpu clock throttled (total events = xxxxxxx)

CPUn: Package temperature above threshold, cpu clock throttled (total events = xxxxxxx)

I realize that this is a hardware warning message (machine check exception, correct me if I am wrong). How do I turn these messages off? Since it doesn't seem to have a detrimental effect of my calculations or my computer (presumably), I don't like it cluttering up my virtual console screen with hundreds of these messages.

Solution 1

You said:

I had cpu cooling (fan) but it went kaput :(

This is why you are having this problem. Your CPU is overheating; these messages are warning you that the system is throttling down in order to prevent permanent physical damage to the system.

I generally expect a Core i7 desktop CPU running under full load to be somewhere around 75-80 degrees Celsius. At around 95-100, thermal throttling kicks in. If it gets significantly above that, the computer will power itself off without warning in an attempt to prevent permanent damage.

The proper thing to do is to replace the CPU fan. Replacement OEM fan/heatsink combos can be had on eBay for less than $20; or you may want to invest a little more cash in something better if you intend to overclock, or run lots of these computations. A better fan may run quieter for the same CPU temperatures, or may cool the CPU more effectively allowing for overclocking, or both.

Ignoring this message, you will eventually destroy your CPU or lose significant amounts of performance or both.

Solution 2

I have no knowledge about possibility for turning off such warnings, but I'd like to point out, that actually - your kernel needs to lower your CPU frequency because of constant overheating. It's not a warning about high temperature 10 degrees before the yellow light would light on. I'd recommend you to try to get better cpu/mainboard cooling - otherwise your hardware can get seriously damaged in not so long period of time.
Although it's not a way to switch off the warnings completely: you can pipe the output of the program to a file and later use less or watch with cat/tail to see the newest results without the warnings.

Solution 3

In order to favorably remove the warning clutter IS to address the hardware component that has failed (the fan) or its related components (heat sink clogged, vents blocked), just like if your SMART data for your hard drive was warning you of bad blocks. Yes you can keep usung the drive but the possability of imminent failure is being given to you to try and avoid catastrophic loss of data. Thus back up and replacment would be the wise advice. So in short you need to address the kaput fan. You are being told that the system is throttled(slowing down, in an attempt to stay running and reducing heat). The warning is that the system has acheived or nearing is max temp threshold and that safegaurd procedures may take over before permanent failure occurs. Or Permanent failure will occur.

I can represent this with a little hummor :)

if y + f > G = 0 M

y = Bear Grylls (he represent you)

f = fish (your mathamatical load to the CPU)

g = Grizzly (Heat build up)

M = chubby sportsman on wheel (your CPU)

When you

Apply

to

Doing the computations..

with no fan or means to disipate the heat...

You will end up with a grizzly with full belly of chubby sportsman and the wheel will stop turning!! You will eventualy have to get a new chubby sportsman to run on the wheel. That can be expensive when a simple fan replacment will fix the problem and allow your CPU to opperate a maximum performance.

Solution 4

In the case of such errors You should install the package :

aptitude install lm-sensors

After that You could check sensors temperature with following command :

sensors

If it's too high (around 100%), You should check Your cooling system.

View more solutions

Comments

[grs]

First of all, why do these messages appear? Do you have a hardware problem? Second, do you have some sort of BIOS protection (unlikely, but just in case). Third, did you install lm_sensors to check at what temperature this happens? And lastly, what produces these messages? I never seen them before but I never had a temperature troubles either.

[Ivan]

I second lm_sensors. It's not bulletproof - it can give some errors in temp meausures - but should indicate any problems when the system is under load.

[dearN]

@grs I have lm_sensors. This happens when the temperatures go beyond 94C.

[dearN]

@Ivan Same answer as that given to grs

[grs]

@drN - 94C is very high temperature for a CPU. I see your CPU fan is gone from your comment below. The best way to fix your problem is to buy new CPU fan. Those are not expensive or are much cheaper than the price you would pay for the whole machine (or even property damage if the machine catches fire, God forbid). Since I am not answering your proper question, I am putting this as a comment. Just spent $20 and forget about the whole thing :)

[terdon]

I don't understand, you're running your computer with no cpu fan and you are wondering why you are getting heating problems? I don't know why it even works at all! Run, and I mean run, to the nearest store and get yourself a fan, or just wait a few days until the thing dies on you. Your choice :).

[dearN]

@terdon, thanks for your emphatic comment. Define cpu fan. My cooling pad died. NOT my cpu fan.

[dearN]

@terdon, my nearest store which might have computer paraphernalia is in the next state literally.. so I guess Ill just run to my nearest web browser and order a cooling pad on amzn! :P

[terdon]

Sorry for the overreaction, I took your comment to mrówa's answer to mean that your actual CPU fan (the one sitting on top of your CPU's heat sink) had died. Hence the urgency :). In any case, you really should not be seeing these kinds of temperatures. Sure, get a new cooling pad but that will just be a stopgap. As everyone else has said, something is seriously wrong here and if it continues it will most certainly have a detrimental effect on your computer, and your battery.

[dearN]

I had cpu cooling (fan) but it went kaput :(

[dearN]

Oh crap! What I mean is the cooling pad driven by USB went kaput. Do you think just by changing the cooling pad, I'd have better results?

[Carl B]

It may help. But, your system still exibits thermal disipation issues. So, your heat sink may be plugged with dust and so on. Your CPUs temp running that hot under load is alarming and eliminating the computers "white flag" of surrender really needs to be addressed by whats causing it. If you are not comfortable about opening it up and blowing it out with some air, a profesional service will be far cheaper than replacing your i7 core or other critical components. :

[dearN]

What do you think an i7's temperatures should rise to whilst solving a complex equation that eats up the RAM and uses up 4 out of the 8 cores available? Is ~90C really too high?

[Carl B]

If it is "cluttering up your virtual console screen with hundreds" of warnings than I am siding with the your system in it's aattempt to tell you that it is too hot.

[Carl B]

@drN - also, asking what it should rise to is speculative. It can go as high as the CPUfan/Heat sink will allow BEFORE the transmission of the Core temp above threshold warnings appear. This temp ranges all over the place depending on the type of i7 you have. Thus, to my comment, I side with the system when it TELLS you it getting too hot.

[Carl B]

The op indicated in other comments that it was not the cpu fan, but rather a cooling pad fan hooked up to the USB. Although you and I agree that there is a thermal issue and addressing the issue is the answer, not eliminating the warning notice. Cheers:)

[Xen2050]

What's a 100% temperature like? my sensors doesn't show percentages, and the high/crit/hyst temps can be edited