Overheating/CPU temp issues
 

I'm currently running SimplyMEPIS (highly modified, now essentially Debian Sid) on an Alienware Area51m. This thing is having some SERIOUS heat issues and it overheats constantly. I feel like I've bought a Ferrari with a microchip in the engine that won't let it go over 30m.p.h.

The computer runs fine when I'm doing normal tasks. If I start running Java apps (e.g. Limewire, Azureus), I can hear the fans start to spin faster. Compiling ANYTHING or running an emulator makes the fan so loud I can barely hear anything else. Try to run a game? Ha! I'm lucky if I get ten minutes before it crashes completely. Usually, it's more like two. My wife has these lovely cold packs (basically a sack of corn or something similar wrapped in fabric) that I can yank out of the freezer and apply to the back; doing so while gaming will often give me an additional 10 - 20 minutes of game time before DEATH.

I thought it might be related to ACPI, but that's all installed and seems to be working. I noticed the modules are being loaded at boot (battery, cpu, and a few others, I don't recall). I also figured dust in the CPU vent might be a problem, but I'm vacuuming that out frequently and I don't see any obstructions to the airflow.

I've tried installing sensor packages (ksensors, hddtemp, etc.), but I'm not getting any readouts. That's strange, because with such a new computer I'd expect the hardware sensors to be functioning. My CPU temp is also strangely locked at a (greyed out) 75 degrees C in ksensors. "acpi -t" returns the same value permanently. I could immerse the entire computer in liquid nitrogen and it would still tell me the CPU was 75 degrees. Obviously, something is wrong there, but I don't know if it's related to the overheating problem or something else entirely.

I'm not going to say that it never overheated with Windows, but I don't recall it ever being this bad. It also seems to be getting progressively worse (the computer is about a year old now), but that could just be my imagination. This all suggests to me that it's something Linux related, and possibly something that I could fiddle around with some settings and fix. I'll get some help when my new Cooler Master Coolpad arrives later on this week, but I fear that will only be a partial solution. I'm trying to get Windows XP installed through qemu (it is not possible for me to work without a small number of Windows apps), but the computer always overheats 10 to 15 minutes into the installation. I'd imagine it would run approximately the same way.

Any help with diagnosis/fixing would be greatly appreciated. Now, if you'll excuse me, I'm going to use the bottom of my computer to cook a steak...

It sounds to me like your CPU just isn't getting any airflow, and I'm wondering if maybe the fan just isn't getting any power, maybe due to a power cable connection that isn't seated securely.

More importantly though, given that your machine is approx 1 year old it may still be under warranty. If so, you may want to turn it in for checkup. I don't think that Linux itself is the cause of the problem; my guess is that it's some sort of underlying hardware issue. Good luck with it.

Lastly, if you aren't getting any readouts from lmsensors, it may be that you just need to manually load a kernel module or two

I would doubt that the CPU isn't getting any airflow. The fan itself is at least working. I can hear it running right now, and I can feel the air coming out of the back. It does make sense, however, that I may not have the necessary modules running or have the sensors configured correctly. I guess I need to do a little more homework on the sensors.

My computer is still under warranty (3 years), but I don't want to return it unless absolutely necessary. Alienware has made clear that they automatically wipe all data from any hard drives that come back to them; no exceptions. They then restore the software to exactly the way it was when it left the factory. As fun as it would be to spend quite some time backing everything up, go for days (if not weeks) without my only work computer, and then have to once again wipe the hard drive and completely set up Linux AGAIN when it finally comes back, I'd rather avoid that if possible.

I'll look into the sensors and see if I can find some more info on that. I know less than nothing about kernel modules, though, so I'll have to study up.

If the fan blow at the same force no matter what, then I suppose it is some sensor -related problem. If fan speed varies, then it is more likely a hardware problem because the fan is actually doing what it is supposed to.

Anyway, you can at least minimize unnecessary heating by using "cpu frequency scaling" kernel modules and an appropriate daemon. These lower the CPU's clockrate under low load and raise it on high load. You can do a "cat /proc/cpuinfo" when doing nothing and again when under heavy load to see if cpu frequency scaling is working.

Kernel modules that you should load depend on your CPU and ACPI version. Www.google.com/linux will help. The simplest and easiest-to-configure daemon is "powernowd". There are others also, for example "cpudynd" and "cpufreqd" - if I don't remember their names incorrectly.

Also check out these sites... they might have howto's on installing and configuring linux on your specific laptop:

www.linux-laptop.net
www.tuxmobil.org

I happen to have the same Notbook and the same problem. First off, any CPU related tools won't work as far as i get it, since the notebook is equipped with a desktop CPU that doesn't support any frequency changes. What I found out myself while disassembling the latop is that the "cooling paste" (dunno if it's the right word) for my CPU just had gone. I cleaned the cpu so far and it got a little better, but since the cooling is poorly made from what I see, without the paste the cooling cannot be too efficient as the copper cooler doesn't really touches the CPU itself as the design is clearly made for use of cooling paste. I'm currently waiting for my order of paste to see if I can fix my "broken" laptop again. One thing i found out is that nice dmesg output I get:

[4300673.251000] CPU0: Temperature above threshold
[4300673.251000] CPU1: Temperature above threshold
[4300673.251000] CPU1: Running in modulated clock mode
[4300673.251000] CPU0: Running in modulated clock mode
[4300678.256000] CPU1: Temperature above threshold
[4300678.256000] CPU0: Temperature above threshold
[4300678.256000] CPU0: Running in modulated clock mode
[4300678.256000] CPU1: Running in modulated clock mode
[4300707.982000] CPU1: Temperature above threshold
[4300707.982000] CPU0: Temperature above threshold
[4300707.982000] CPU0: Running in modulated clock mode
[4300707.982000] CPU1: Running in modulated clock mode
[4300728.207000] CPU0: Temperature above threshold
[4300728.207000] CPU1: Temperature above threshold
[4300728.207000] CPU1: Running in modulated clock mode
[4300728.207000] CPU0: Running in modulated clock mode
[4300858.305000] CPU1: Temperature above threshold
[4300858.305000] CPU0: Temperature above threshold
[4300858.305000] CPU0: Running in modulated clock mode
[4300858.305000] CPU1: Running in modulated clock mode


Where you can see clearly (when working with the latop you also notice the fan increasing like hell) when the point of overheating is reached. Usually my system works well then given I use programs that don't intensivly need the cpu like text processing, etc. But if I try games with more intense graphics or start a compilation of a program with gcc for example it doesn't take until i get freezes with a following shutdown and the impossibility to boot up the laptop until he got a decent amount of time for cooling down and/or I get error messages mostly from kswap telling me that several CPU registers are not the way they are supposed to do, thus causing a crash which mostly can be fixed, but I get advised to reboot anyways, which in turn results in a several minute lockdown of the whole system (usually I need to wait for about 15mins until I can boot up the PC again).

I am no expert at all when it comes to that level of hardware and linux, even if I maybe have some above-the-average knowledge, so all of the above is just my reasoning to find the cause for the error and may be wrong at all, which I actually don't hope ;)

Thanks for the update. I'm glad to know I'm not the only one having this problem.

My Coolpad has arrived, and it's a godsend. It hasn't totally eliminated heat-related crashes, but they're extremely rare now. Before, any movies or games could prompt a shutdown, and simply leaving the computer on for an extended period of time (more than 1 day) would often cause enough heat buildup to crash the system. Ever since I got the Coolpad, I can leave my computer on as long as I want and have no problems. I usually only have crash problems now when I'm doing extremely CPU-intensive work. Even then, it seems to only crash because of ridiculously excessive heat building up in a matter of seconds, too fast for the Coolpad to dissipate.

I don't think I'll be recommending anyone to purchase another Alienware laptop until they fix this. I just don't get how they can build and market these computers for gamers. It's a bit difficult for a gamer to play anything when the computer crashes after five minutes. They really need to figure out a better way to dissipate heat.

AFAIK the Area 51m (S-51m in my case) isn't sold anymore, probably because of the facts we have to deal with. But the current models are very similar and I personally have no clue if they have the same problems of not. Anyways glad I was of maybe a little help ;)

I have the same laptop with exactly the same problem overheating anytime the cpu is taxed and crash after 5 min. The cpu cooler uses a heat stick to transfer heat from the cpu to the fan my assumption was there was a leak in the stick, the refridgerant had leaked out rendering the stick a hollow peice of copper. Turns out that isn't the case the cooling sytem works fine it's power management and frequency cycling. A good buddy of mine discovered and corrected the problem on his own and passed the solution on to me. Unfortunatly he's a windows guy and knows little to nothing about Linux. Seems in Control Panel in Windows there is a Power Options setting that Windblows defaults to and another one the Acer likes. Windows defaults to 'Max Battery' which overheats the crap out of these units however if its set to 'Laptop' the unit runs fine can do very long stretches (20+) hours of serious gaming yet never crash. Now I don't know how this can be fixed in Linux but I'm pretty darn sure if a simple setting change can solve the problem in Winblows then it should be just as easy with Linux.

Hi guys - I have an alienware that crashes due to heat too.
A cheap/homebrew method is this: prop up the back end of the lappy with two small peices of wood (or whatever you can get your hands on) get a small table fan (cheap ass ones from the £1/$2 store) and point it at the back where the fan is. I live in japan and the summers here are as hot as hell - but when i put the table fan at the back - it never crashed once - its seems that the fan cant suck in enough cold air - the gap is too small.
if it does overheat - the quick way to cool it down again is by using 'cool pads' you know the ones for muscle sores or burns (you put them in the freezer). chuck one underneath or on top of the keyboard and it a minute you lappy will be cool and will last for hours.
I'm not a cheap ass - the only reason that I do this is because i live in middle of Japanese no-where, i have no computer stores or hardware storess so i gotta use what i can.
Its not a permanant remedy, but its certainly a quick fix. I also hear blowing "air in a can" through the vents to unclog the dust works well too.
Just my bit!
CF

It seems to be a serious hardware issue. Mine's still under warranty, so I called Alienware. They had me run a CPU stress test program (which I was happy to see runs fine with Crossover Office!), and that crashed the computer in less than three minutes. They're suspecting it's a heatsink/fan issue (which sounds logical), so they're supposed to be sending me a replacement and installing it for me. If any of you have this same issue and still have warranty time left, it might not hurt to call.

I'll check back once the parts have been replaced and let you know if it did the trick.