So what *does* make a PC go catatonic?

[jducoeur]

Here's a semi-idle question for those with more hardware-fu than myself:

My laptop has a problem, which it's had for quite some time. (I really should have exchanged the bloody thing when it was still new enough for that to be an option; now, I just live with it.) Once in a while, for no particularly apparent reason, it simply locks up -- hard. There's no BSOD, nor the runaway fan that would indicate an out-of-control process: it just stops responding to anything. I don't think it's a software problem, because the mouse goes completely non-responsive as well, and it's pretty hard to crash Windows badly enough to kill the mouse. (Almost everything else in Windows is fairly easy to destroy, but the mouse operates way down deep in the kernel.)

Anyone have any ideas what would cause such behaviour? The screen stays active where it last was, but I can't *do* anything -- even the three-finger salute has no effect, so I have to hard-boot when this happens. I'm mildly resigned to living with it until I decide to get my next laptop (which, yes, might actually be a Mac: my PC-centrism is weakening in this new Net-centric world), but any suggestions would be welcome. It's a real PITA when I'm in the middle of editing something and it does this...

Comments

[laurion]

A lock up that hard is usually very low level. It could be bad memory. Have you added any extra memory at any point?

Run a Mem386 test?

[jducoeur]

No, and no. Probably not a bad idea, but I haven't tried the test before. How does one run it?

[laurion]

http://www.memtest.org/

misspoke earlier. mem386 was a DOS TSR, heh. memtest86+ is what you want. It can be installed and run from windows, but it is also part of many linux-based CD recovery systems such as http://www.sysresccd.org/.

[laurion]

You might look at http://trinityhome.org/Home/blog.php?front_id=15 , a distro aimed at helping out windows based systems.

[jducoeur]

Lovely -- I'll try this out. Thanks!

[dsrtao]

Simple test: go burn a LiveCD distribution -- say, Ubuntu (www.ubuntu.com).
Run it, get it to a desktop, and leave something running overnight.

If it doesn't die, it's probably something Windows is doing. If it does die, it's probably hardware.

[g-w-s.livejournal.com]

Both are good ideas. You do not yet know if it is hardware or software related.

It is most likely related to the memory, the processor, or possibly even instability in your power supply.

Try memtest86. That will give you confidence there is not a memory problem. If it does crash, that will also give you confidence it's not a windows problem. I say this because it will run forever without requiring any fancy windows drivers, and any crashes at that point are tied very closely to the hardware.

Since it's a laptop, there's also a good chance that it may be overheating. You can accelerate an overheating problem by keeping the processor very busy (linux users: while true; do make bzImage; make clean; done) (windows users: play a cpu-intensive game).

You can also try keeping the laptop cooler - make sure there is adequate ventilation underneath it; for desktops, I usually point a fan at the open case to check this problem, but you can find other ways of keeping it cool.

If you notice temperature affects the stability of the laptop, that definitely points towards a thermal-related problem. This is typically the processor, though it could be other things. For example, when my video card would overheat, I would see strange pixels on the screen or strange artifacts in 3d games - eventually it would lock up the bus and hang the system, but not without obvious warning.

[jducoeur]

I've thought about the heat problem. It's possible, but so far I haven't demonstrated that it's so -- running a thermometer program hasn't shown any clear evidence. That is, it doesn't seem *unusually* hot when the lockup occurs, but I don't have enough data points yet.

I'll try the memory test: so far, that seems the likeliest-sounding bet. Thanks!

[jducoeur]

Actually, while that makes some sense, I don't know that that'll do it. The lockup is *very* unpredictable. Sometimes I'll go for weeks without it happening; sometimes it'll happen two or three times in a day, and I'll be ready to hurl the damned machine across the room. So while this test could provide positive proof that the problem is hardware if it *does* lock up, it won't provide negative proof if it doesn't.

(My suspicion is that laurion's guess that it's memory-related is correct, actually. Now that I think about it, it has only ever happened when Firefox is running, and never when I'm only running iTunes. Which is significant because Firefox is a *dreadful* memory hog, and likely to stress the memory more than most things. Which implies that a simple overnight test might not do it. What I probably need is a serious memory test.)

Still, an interesting idea, and worth keeping in the back of my mind. I hadn't realized that there were straightforward run-from-CD distributions. That's worth knowing about; thanks!

[qedrakmar.livejournal.com]

My first thought is thermal... that's usually the most unpredictable factor. It might not even be the temp of the CPU, so much as the temp of, say, a voltage regulator somewhere that causes bit streams to cross in the bus followed by a kernel panic, or something equally random. So, you might try something like putting it on a cool metal desk in front of a window, and see if that helps the stability. (How often does this happen, anyway?)

On another thought, if the RAM comes up clean... how much RAM is in it? It could still be a billion different things, but another likely one could be HD issues, in the region of your swap.

Another... though this is kind of a long shot, do you have issues soft rebooting it that force you to do a hard reset? (If so, I know what that is...)

[jducoeur]

(How often does this happen, anyway?)

"Intermittently", unfortunately. Often enough to be annoying, but not often enough to make experimentation straightforward. Maybe once every few days, typically.

another likely one could be HD issues, in the region of your swap.

Interesting. Pain in the ass if so, but worth keeping in mind as a possibility.

do you have issues soft rebooting it that force you to do a hard reset? (If so, I know what that is...)

Ah, good point. No, I don't have to do the *really* hard reset. Pressing the power button for five seconds suffices to turn the machine off.