Hi!

Don't really know if this is the right place to ask about this... but here it comes.

Some months ago I began to notice a weird problem with one of my computers. This box is a dual boot with Win XP and a bunch of linux distros. I rarely use windows, but it so happened to be that I was playing a couple of games using XP at the time. It was then I began to notice a slightly odd behavior. After playing (3D games, rather system-intensive) for a while running XP, I restarted to boot linux. Some times it would boot fine, but some times it would stop on BIOS check and tell me something like "Hardware Monitor Test was not successful, please revise your BIOS configuration". Well, I looked at my setup, but did not find anything unusual or wrong. So I just restarted, and all went fine after that. This happened several times.

But I did not have any major problems until today. Today my computer hang while I was running linux. I was forced to use the power button to restart the pc. Now I got the weird BIOS message again. This time around though, I went to the BIOS Hardware Monitor and guess what I found? My CPU temperature was at about 90C.

Since this happened my computer has hanged two more times. I have tried to use lm-sensors to get a measure of the core temperature run-time. That gave me some weird results:

running X server nad nothing more: 60-65C
during heavy workload: 70-75C
Maximum fan speed in both situations (4500 rpm)

That is rather high, though it is not yet critical. Even so, it hanged when I was measuring it under load and the temperature at that time was "only" 72C.

The worst thing is that *everything* seems normal. While both power supply and CPU are really hot (I tried to touch them with power on) their fans seem all right, and the GPU is not too hot either. System log seems OK too...

Does anyone know what this might be and what could have caused this?
Any help is greatly appreciated! Thanks in advance!


My specs. This computer is about 3 years old.
ASUS A7V600 mobo
512 mb RAM
nVidia FX5700 256mb
AMD Athlon XP 2500+ 1.8GHz
Custom-compiled 2.6.21 kernel

Btw, can this be caused by a misconfigured kernel?

edit: If I let the pc cool down, then turn it on and watch the temperature in BIOS Hardware Monitor, it slowly rises to about 83C, then stops (or at least slows down considerably).

Those temps seem quite hot to me. However, I'm not sure what that generation of CPU "should" be running at temperature wise.

As a comparison, I have a 3.2Ghz P4 Prescott (aka pressed hot) and during extended (hours of constant 100% usage) my temp never goes above 55C.

I would definately clean out your cpu heatsink/fan using canned air. The fins are easily clogged by dust etc that may not clearly visible due to the design. If that doesn't help, perhaps removing the heatsink/fan give the it a good cleaning, remove the old thermal paste from them fan and processor. Then apply new thermal paste and put it back together.

If your power supply is hot, perhaps it is slightly overloaded or it may just be picking up the heat from the inside of the case.

If you have other case fans, check them to make sure they are functioning up to par so you get enough fresh air into the case. If you have other temp sensors on your MB those can help you as well. If everything is heating up bad, perhaps it's an air-flow issue. I fyou have an air filter on the case, make sure it's clean.

Sorry for the rambling, hope this helps.

Normal CPU temps usually should be in the 40's or 50's C, with occasional higher spikes during heavy loads, but 90 C is way too hot.

I would guess that either your CPU fan isn't seated correctly, or that it's not working properly. What you're describing is a classic overheating issue though, and docalton's suggestions are exactly what I'd also suggest: confirm all fans are working, esp. the CPU fan, blow out the dust, and make certain that the CPU fan is seated properly

Thank you very much for your answers!

I have actually cleaned the "insides" of this box a month ago so I had kind of ruled out that possibility already. Though when I took a look at it now, I found out that I managed to clean only the outside of the fan, and somehow totally forgot about the heatsink and CPU fan's back side.

It was rather dusty, so I cleaned everything out.

Although temperature is now slightly lower (it used to be 63C, now 59C), I still think it's a bit too hot. I haven't had any crashes since the cleaning, but I am not sure whether it's pure luck or the result of the cleaning.

I will try to put some load on the system and see how it does.

Again, thanks a lot for your suggestions!

When I had a Athlon 2600 60 degrees was nothing special.Booting put it up there already.
Dunno about the max temp it went because I don't monitor temps but that thing was happily compiling for days at 100% CPU.

Now 90 degrees max seems a bit high.Do you got a CPU fan that has different speeds?
It might run slow but doesn't speed up when needed.

Also check if you got some funny stuff in the BIOS enabled (like speed control for the fan).If there is something just put the fan on max.

Thanks for the info, crashmeister I have never measured min/max temperature of my core before, so I don't have anything to compare it with anything either actually.

I also thought about that with fan speed. The problem is though that I have this BIOS option which says "Q-Fan control". The annoying part is that it ONLY controls the fan speed if the tepmerature is low, but not if the temperature is over some threshold which cannot be changed by the user. So although it is set on full speed, the fan speed controls itself (because seemingly the temperature is over the threshold all the time). Thus, what you get is a fan which does not run at full speed on higher temps, but instead scales may be 500-1000rpm down.

Now I have run some testing under heavy load, and well, yeah... After the cleaning the overall temperature has gone down 5C or so. And this time around it did not crash either.

Another weird aspect of this issue is that it seems like there might be something wrong with lm-sensors' output. First of all temperatures for mobo and CPU are swapped. And then there seems to be some kind of problem with reading fan speed. Sometimes it just drops to 0, though I can clearly see that the fan is rotating. This in turn indicates that the output I'm getting may not be the correct one.

Another interesting fact is that like I said earlier, the CPU temperature seems to be reported higher by BIOS than by lm-sensors.

So I am really confused about this whole issue. May be you guys can tell me what is wrong (if anything). Should I may be update the BIOS? Is this a kernel configuration issue?

Forget about lmsenors and the likes (M$ or Linux based) - those apps can't get it right in M$ and Linux.

If the BIOS gives you hickups -as in your case- thats another ballgame.Then you better go after what is wrong.You can try a BIOS update but that can screw up things even more - better read the changelogs.

If your fan runs at 500-1000 you better turn Q-fan off.Somewhere between 1000 idle and 2500 under load sounds more like it.

Does your bios setting allow for more control. I can step the fan on both my desktops with x2 and x4 settings.

As far as I can see, unfortunately it does not. I have the earlier mentioned "Q-fan control" setting, which I have now set to "Full Speed", but that is of no use at all since this setting is overridden by the system when temperature gets high enough. The problem is though that it seems that it turns the fan speed DOWN.

So what happens is that when I start up at about 60C, the system has already passed the threshold of manual control and makes use of a built-in automatic fan speed adjustment. The problem is that it does not consider the temperature too high, and pushes the speed down to about 3000-3500rpm. If I had had control over it, I would have set it to max (4500rpm) all the time, no matter what temperature is. But unfortunately I am not given any option to do so in BIOS my mobo uses (or at least none that I can see).

I think I was a bit unclear here. The fan does not go down *TO* 500 - 1000 rpm, it goes down *WITH* 500-1000rpm. That is, the maximum fan speed of my CPU fan is 4500rpm and right now it's about ~3200 +/- 200 rpm. So it has turned it down a little over 1000rpm if you compare it to the maximum speed.

As for the Q-Fan, well I am not sure wether it will help to turn it off. Take a look at this pic, which probably explain it better than I can: http://www.asus.com/999/html/events/...tures/qfan.JPG

When Q-fan is ON, the fan rotation speed will follow the red line. When I turn it to "Full Speed", the fan speed will follow the green line.

In either case it won't matter, because my temperature is already over 60C; while it has to be < 60C or so for the Q-fan to have any effect at all.

Or am I misunderstanding something here?

Probably obvious but need to rule it out hehhe. Do you have multiple connectors on your mb (board and CPU fan connectors), and have you verified the cpu fan is connected to the correct one. ( I have multiple and would guess the same result if the power connector was on the wrong port i.e. mb temp controlling cpu fan and vice versa would result.) Although that wouldn't explain the onset of the problem.

I think there is only one CPU fan connector on the mb. And as far as I can see it is properly connected.


Well, although this probably is an overheating problem, right now I am unsure what could *possibly* be causing it at all. So all suggestions are mostly welcome So, thank you very much!

Try a BIOS update if there is one available.other than that flush the BIOS.
I didn't tell youthat in case nothing works after that :-)

You can change that if you want them displayed properly in the /etc/sensors.conf file you would need to switch around the lines like these in the section for your sensor chip.

For the heating issue did you reinstall with new thermal paste/grease this eventually wears out over time maybe upgrade to a better heat sink/fan as well.

LOL


Well, no I haven't. I have to order it after this weekend is over or something. Thanks for reminding me about it though


Yeah, that might help and won't be too expensive either, I gather. The thing is that I want to get this pc fixed as cheap as possible (since it's getting old and since it's a rather cheap build and all that). And if it breaks after that, well... then it is its time to be disposed of

Thanks

Though I am still not sure why I am getting 0 fan speed some times...