bypassing thermal protection shutdown of a system from within Linux/Ubuntu

[Turbonium]

Is this possible?

The reason I ask is this thread. Basically, the system's thermal diode has gone faulty and gives bad temperature readings (it thinks it's overheating when it's not). I want the system to ignore the "overheating" and just stay running.

 

[Virgorising]

Is this possible?

The reason I ask is this thread. Basically, the system's thermal diode has gone faulty and gives bad temperature readings (it thinks it's overheating when it's not). I want the system to ignore the "overheating" and just stay running.

How did you determine the diode is faulty and the temps are normal? Yes, I did try to read the thread you linked us to, tho some of it was beyond my current level.

If a system is venting very hot air....while obviously true, were all that trapped, it would be the coup de gras.....just that yr system is venting SUCH hot air I think is cause for more exploration inside. Yes, I get you do not like going here.

A human with a very high fever, also vents very hot air. But that....is a symptom, and inevitable result of something amiss within the organism, yes?

 

[Turbonium]

Check out the thread. The problem is that the diode thinks the CPU is a lot hotter than it actually is, so it tells the system "You're overheating, you need to shutdown to protect yourself".

To use your analogy: imagine someone with a strange health problem where their body temperature is normal, but they feel very warm and think they are overheating. It's a false reading (off topic, but I think some thyroid problems can result in this sort of thing).

Anyway, at this point, I'm about 90% certain the diode is to blame, hence why I want to force the system to stay on.

 

[Virgorising]

Check out the thread. The problem is that the diode thinks the CPU is a lot hotter than it actually is, so it tells the system "You're overheating, you need to shutdown to protect yourself".

To use your analogy: imagine someone with a strange health problem where their body temperature is normal, but they feel very warm and think they are overheating. It's a false reading (off topic, but I think some thyroid problems can result in this sort of thing).

Anyway, at this point, I'm about 90% certain the diode is to blame, hence why I want to force the system to stay on.

OK but, given heat is the ultimate enemy of this technology, I, for one, would not invest in your almost 90% with alacrity.

Re your trying to employ the analogy to confirm your take....if the being in question FEELS VERY WARM AND IS VENTING VERY HOT AIR.....that is not a bad "diode." That is a veritable symptom to take seriously....most often of the body doing what it has miraculously evolved over millenia to do: rev up to fend off some infection or other issue. The little factory, say, of white blood cells goes into overdrive.

In yr situation, I would hazard, the high temps owing to some rare thyroid issue is the Zebra. More likely, there are horses here somewhere.

Vital signs are monitored by professionals for reasons.

I, for one, would not FORCE ANYTHING. I would do more troubleshooting and delving.
__________________________________________________
How do you know, given, you are not crazy about going inside the system (nothing to apologize for, i know tons of people who feel the same), your heatsink is properly seated? How do you know your fan is working? Maybe you need to remove old thermal paste and replace it with new? There are many possibilities here, which I think should be considered before trying to force or circumvent anything.

And I don't know even a percentage of what many here do!!!

 

[Ketchup]

The shutdowns you are talking about are usually controlled in the BIOS, so you should be able to stop the shutdowns from there.

 

[Virgorising]

I just found this....for whatever it may be worth:

http://superuser.com/questions/55632/cpu-temperature-sensor-wrong

 

[Turbonium]

I'm going to try reseating the CPU and the HSF, but if that doesn't work, I'm going to conclude it's likely the diode.

Again: it was working perfectly for 2 months at 100% CPU load, and then suddenly it "broke" and started overheating. It's not like the CPU, HSF, or case position moved. If anything, the ambient temperature went up a bit in my apartment, but that's about it. Rather confusing, really.

The thing that really has me thinking it's the diode is the fact that it reaches Tjmax from a cold boot. Usually, a CPU takes at least a few minutes to heat up to that level afaik (and it's not like booting up into BIOS has the CPU at 100% load). Someone correct me if I'm wrong.

 

[Virgorising]

I'm going to try reseating the CPU and the HSF, but if that doesn't work, I'm going to conclude it's likely the diode.

Again: it was working perfectly for 2 months at 100% CPU load, and then suddenly it "broke" and started overheating. It's not like the CPU, HSF, or case position moved. If anything, the ambient temperature went up a bit in my apartment, but that's about it. Rather confusing, really.

The thing that really has me thinking it's the diode is the fact that it reaches Tjmax from a cold boot. Usually, a CPU takes at least a few minutes to heat up to that level afaik (and it's not like booting up into BIOS has the CPU at 100% load). Someone correct me if I'm wrong.

Hi, Again....

Instead of doing the work I need to finish, I have been delving into this issue.:\ You last point above is a good one, cause it is logical.

But, I still would not take a chance on trying to circumvent the shutdowns via the bios, until you make absolutely sure the system is not, in fact, overheating. Just my take were I having this issue.

Plus, in the rare, it seems to me now, but not impossible event sensors go bad, is that likely to happen in the first two months?

I just found this new one:
http://www.dslreports.com/forum/r26217558-motherboard-heat-build-up

Even if there might be too much, or badly applied thermal paste, the metal can not make contact with the metal properly. I still get nervous doing that, because of that.

 

[Turbonium]

Hi, Again....

Instead of doing the work I need to finish, I have been delving into this issue.:\ You last point above is a good one, cause it is logical.

But, I still would not take a chance on circumvention the shutdowns via the bios, until you make absolutely sure the system is not, in fact, overheating. Just my take were I having this issue.

Plus, in the rare, it seems to me now, but not impossible event sensors go bad, is that likely to happen in the first two months?

I just found this new one:
http://www.dslreports.com/forum/r26217558-motherboard-heat-build-up

Even if there might be too much, or badly applied thermal paste, the metal can not make contact with the metal properly. I still get nervous doing that, because of that.

Hey. Thanks a lot, but you really should go do your work!

Also: when I applied the HSF, it was a stock Intel one with the thermal pad pre-applied, meaning I couldn't have made any errors applying any thermal paste.

 

[Ketchup]

I would really like to see you temps on an app like core temp or CPUID hardware monitor.

 

[Ketchup]

I would really like to see you temps on an app like core temp or CPUID HW monitor.

 

[Turbonium]

I would really like to see you temps on an app like core temp or CPUID hardware monitor.

I'm running Linux. :/

 

[Ketchup]

So you can't run things in Linux?

http://superuser.com/questions/25176/how-can-i-monitor-the-cpu-temperature-under-linux

What distro?

 

[Turbonium]

So you can't run things in Linux?

http://superuser.com/questions/25176/how-can-i-monitor-the-cpu-temperature-under-linux

What distro?

Check the link in the OP. I posted temps. Link. Am I missing something?

Ubuntu LTS, latest

 

[Ketchup]

Check the link in the OP. I posted temps. Link. Am I missing something?

Ubuntu LTS, latest

I see three temps on a quad core processor. Weird. But you've got to work with what you have I guess.

Saw the link to the case also. I just don't know that I would put a quad of that generation in a case with such little airflow.

 

[Turbonium]

I see three temps on a quad core processor. Weird. But you've got to work with what you have I guess.

Why is that weird? I thought one (the highest) temp is the CPU, and the other two are the motherboard and chassis or something. Does each core have a different readout? Although come to think of it, my dual core T9500 has 2 temp readouts... perhaps it's a bug or limitation in lm-sensors?

Saw the link to the case also. I just don't know that I would put a quad of that generation in a case with such little airflow.

Well, it's been fine (and is fine) in the other box. It's literally crunching away right now, and has been for 2 months.

I know what you mean though. I might end up installing an exhaust fan in each box at the top, since it has room for it. That might help things in general.

 

[Ketchup]

My Q6600, i5's, and i7 all show(ed) 4 temps - 1 for each core. But it's linux and an Intel board, so I am not too surprised things are different.

Did you find a setting in the BIOS to stop the shutdowns?

 

[Turbonium]

My Q6600, i5's, and i7 all show(ed) 4 temps - 1 for each core. But it's linux and an Intel board, so I am not too surprised things are different.

Did you find a setting in the BIOS to stop the shutdowns?

Nope. No options given. I might try looking into updating to the latest BIOS, but that seems drastic, and I always run the risk of essentially bricking the mobo.

Can I bypass the thermal trip somehow though? I'm not necessarily going to do it even if it's possible, but I'm really curious. I'm guessing no, since if the BIOS thinks it's hot, that's game over.

 

[VirtualLarry]

(and it's not like booting up into BIOS has the CPU at 100% load).

Generally, it does. Maybe not 100% load like an FPU stress-test program link LinX, but very nearly a 100% integer operation load.

 

[Virgorising]

Generally, it does. Maybe not 100% load like an FPU stress-test program link LinX, but very nearly a 100% integer operation load.

Wow. What a wicked important thing to know!!!!

I had no clue. But now, I have a clue.:thumbsup:

 

[Turbonium]

Yea, might as well let this thread die now (my system was, in fact, overheating... see other thread).

Unless, of course, someone knows the answer to my original question, for the sake of curiosity and general knowledge.

 

[Virgorising]

Yea, might as well let this thread die now (my system was, in fact, overheating... see other thread).

Unless, of course, someone knows the answer to my original question, for the sake of curiosity and general knowledge.

Well, obviously, I am not surprised, but also soooo happy/RELIEVED you DID DELVE and honored the symptoms as opposed to trying to force anything, given that would have been a total coup de gras.

I, and bet everyone else as well, would love to know how you made the final diagnosis, and what, precisely caused the overheating????

Another case of it's almost never zebra; it's almost always horses.

Congrats!
______________________________________________
Edit: I am unclear why you say flashing yr bios to update is, in some way "risky", esp, given it is necessary, and, I always thought, just normal. That was almost the first thing I did in this new/used system after I set it up....then updated all the drivers. I never ever thought this was something dangerous, nor, at least in my, somewhat limited experience has it ever been.

 

[Turbonium]

I, and bet everyone else as well, would love to know how you made the final diagnosis, and what, precisely caused the overheating????

Defective HSFs (and they are relatively inefficient Intel ones to begin with, so a defective one with one or more faulty pins/clips would just make things even worse).

Just check out the other thread and let this one die. But thanks for your well wishes.

 

[Virgorising]

Defective HSFs (and they are relatively inefficient Intel ones to begin with, so a defective one with one or more faulty pins/clips would just make things even worse).

Just check out the other thread and let this one die. But thanks for your well wishes.

Interesting. I did not know stock fans which are packaged with Intel chips were that sleazy.

Beyond "good wishes," I remain relieved you slowed down and investigated and did not do what, at first you were hell bent on doing.

 

[Turbonium]

Interesting. I did not know stock fans which are packaged with Intel chips were that sleazy.

Beyond "good wishes," I remain relieved you slowed down and investigated and did not do what, at first you were hell bent on doing.

I got them from an eBay vendor, but they were listed as brand new, and appeared to be when I got them.

I've never actually had problems with stock Intel coolers that I've gotten with a retail Intel processor (two instances of S478 gen coolers for Pentium 4 chips, and two instances of LGA1155 coolers for i5 chips... all four were fine).

It makes me wonder a bit about the eBay vendor.