Celestia Forums

Hi, guys and gals,

My desktop computer at home started to misbehave some months ago. It started to reboot or power down by itself. On ocasions it wouldn't reatart immediately after shutting down, but required some (cooling?) time before reacting to the Big Green Button. After careful monitoring I suspected an overheating problem. I replaced thermal grease (yes, I know how to do it properly), cleaned the cabinet, added a fan, left it open... Sometimes it solved the problem for a few days, then it started to reboot again. Even with no task running, just the desktop.

So I took it to a technician, who replaced the power source (I concede that it had started to make a screeching noise, not permanently though). The cabinet is a good quality one, silent, powerful. The new power supply is the same brand as the original. After I brought it home, it started to reboot by itself again.

Today I had an epifany and removed the graphics card, and switched to on-board graphics. The computer has been running flawlessly, no rebooting, for 5 hours. The removed card is an nVidia GF 6200LE (AGP, VGA+SV+DVI, passively cooled; I don't remember the manufacturer, and it's not labelled). I'm downgraded to Intel Extreme Graphics now (such over-rated name!), very bad for Celestia, sniff.

Is it possible that a malfunctioning video card made the machine reboot or shutdown?

Guillermo

More system details (not new, custom made to precise specifications):
* Intel P4 Prescott @3GHz, on an Intel D865GBF board, 2GB DDR (flawless, correctly mounted in dual channel).
* 2 SATA hard discs (no errors, no SMART warnings), DVD burner (hardly used), FD drive (what for!?)
* XP Pro SP3.
* No, there are no viruses, microbes or other protozoa. Drivers are as up to date as they can be, everything is nice and tight on the software front.

abramson wrote:Hi, guys and gals,

My desktop computer at home started to misbehave some months ago. It started to reboot or power down by itself. On ocasions it wouldn't reatart immediately after shutting down, but required some (cooling?) time before reacting to the Big Green Button. After careful monitoring I suspected an overheating problem. I replaced thermal grease (yes, I know how to do it properly), cleaned the cabinet, added a fan, left it open... Sometimes it solved the problem for a few days, then it started to reboot again. Even with no task running, just the desktop.

So I took it to a technician, who replaced the power source (I concede that it had started to make a screeching noise, not permanently though). The cabinet is a good quality one, silent, powerful. The new power supply is the same brand as the original. After I brought it home, it started to reboot by itself again.

Today I had an epifany and removed the graphics card, and switched to on-board graphics. The computer has been running flawlessly, no rebooting, for 5 hours. The removed card is an nVidia GF 6200LE (AGP, VGA+SV+DVI, passively cooled; I don't remember the manufacturer, and it's not labelled). I'm downgraded to Intel Extreme Graphics now (such over-rated name!), very bad for Celestia, sniff.

Is it possible that a malfunctioning video card made the machine reboot or shutdown?

Guillermo

More system details (not new, custom made to precise specifications):
* Intel P4 Prescott @3GHz, on an Intel D865GBF board, 2GB DDR (flawless, correctly mounted in dual channel).
* 2 SATA hard discs (no errors, no SMART warnings), DVD burner (hardly used), FD drive (what for!?)
* XP Pro SP3.
* No, there are no viruses, microbes or other protozoa. Drivers are as up to date as they can be, everything is nice and tight on the software front.

Guillermo,

this is of course a very awkward situation. My first guess would have been a memory fault, but it seems you have checked it ALL!? Did you really do a complete memory scan!?? Next, the autostart utilities are often causing such problems sometimes due to malware (not spotted by virus software).

What happens if you do a safe boot (accessible after hitting F8 right after the start) ??
If you don't get a reboot after a safe boot, then there are good chances that the culprit is one of the autostart utilities (which mostly are installed without you noting it).

Then, next in the strategy, you type Run and enter 'msconfig' which leads you to the panel where you can selectively deactivate various classes of autostart programs. By proceeding step-by-step you should soon be able to locate the cause of the problem.

There are further useful tools in this game, but perhaps you give me first some feedback.

Fridger

abramson wrote:Is it possible that a malfunctioning video card made the machine reboot or shutdown?

That's entirely possible. A friend of mine was having similar reboot problems. When he replaced his graphics card, the problem went away.

Edit: I'm not say that it's the actual cause. Just that it's possible.

Is it possible that a malfunctioning video card made the machine reboot or shutdown?

It just happened to me 2 weeks ago with my Dell notebook. But the symptoms were much more dramatic. The screen suddenly disintegrated into stripe segments and text fragments all ending in a blue screen. There is one test: if you hit F2 right after boot start you get to the bios. If there you also get a reboot then it may well be the graphics card, since all the rest has not yet booted.

Did you replace drivers recently? Notably harddisk drivers??

Fridger

I had a similar experience with a passively cooled (i.e. no fan) Nvidia graphics card on my computer at work a few years ago. All the problems went away when it was replaced by an equivalent version which had a fan. I believe both were 4000 series cards, although I'm not sure. I do recall that the passively cooled one was made by EVGA, while the actively cooled one was made by ASUS. (My current computer at work has an actively cooled Quadro FX 550. I don't know who made it: it came in a preconfigured Dell.)

Many Nvidia graphics chips run hot, and get extremely hot when they're drawing 3D graphics. Overheating will cause them to do strange things. Can you provide a fan which blows directly on the card? Of course, running too hot for a long time can cause permanent damage, so it may be too late for that card.

This is something else. Your problems sound a lot like the problems I had a month or so ago. My computer is still limping but has been damaged. At first, I also suspected an overheating problem, cleaned it, replaced the power supply, etc. Problem is when overheating is the cause, these newer system boards and cpu's seem more vulnerable. As with my system the damage may have been done although I suspect my damage was more than just overheating. I think it had something to do with the power supply.

Today I had an epifany and removed the graphics card, and switched to on-board graphics. The computer has been running flawlessly, no rebooting, for 5 hours. The removed card is an nVidia GF 6200LE (AGP, VGA+SV+DVI, passively cooled; I don't remember the manufacturer, and it's not labelled). I'm downgraded to Intel Extreme Graphics now (such over-rated name!), very bad for Celestia,

This is possible but it may also be possible that the bus the card occupied may have been damaged by overheating. How bad did it overheat? Were the heatsinks and fans very dusty? Replacing the card would depend on how dirty the inside was which would give a better idea of how much overheating was involved.
Sorry, I can't really make a recommendation here. I'm looking to replace everything. There is always a chance that adding new hardware to a system with problems could damage the new hardware too.
cartrite

My card in my Dell Latitude (D620) was a NVIDIA Quadro (NVS 110) which also got very hot before dieing. I never had any reboots, though. Only the "screen decay" ending in the blue screen with stripes...

Incidentally, at which point does the reboot take place?? During the boot sequence or after it? In any case a safe boot should be tried first.

Fridger

Many thanks for all the answers. I had already checked some of the recommendations (but was lazy to report everything), but still some others will be useful.

For example, when I added the fan, I didn't make it blow directly on the GPU difusor, but made it take in fresh air from the front instead. Eventually I removed it because it was noisy and didn't solve the problem (ball bearing and noisy, can you believe it; it was nicely colored orange, though).

The F2 and wait is a good idea, Fridger. I will try it (if I reinstall the card).

The startup configuration has already been streamlined. There is always a possibility that I skiped something.

The overheating was never very bad. At the beginning, when the symptoms started and I began to monitor, I noticed a peak of around 70 degrees (CPU) just before the reboot. 70 deg is not much according to Intel specifications for my processor. But there was a correlation with the reboots, so I connected both causally. But then, in several of my failed attempts, the temperature was much reduced: open cabinet, throttling-down of the processor, etc. I started to put everything in doubt, bad T sensors, expired thermal grease (only Argentine made grease is available in Bariloche, no silver wonders here)...

Seven hours without problems now, with the video card sitting on my desk... (I stayed at home today, we had three days of programmed power outage (10 hours each) in almost the whole city, saving my home but affecting my office. Why don't they do it at night, so I can use it for astrophotography?)

Guillermo

t00fri wrote:Incidentally, at which point does the reboot take place?? During the boot sequence or after it? In any case a safe boot should be tried first.

Sometimes after a couple of hours of web browsing (flash-intensive) there would be a reboot and one could just continue using it without problem.

Some other times it would shutdown and it wouldn't allow to turn it on again for several minutes (these modern power suplies are getting a little too smart!).

Still other times it would enter a cycle of several incomplete reboots (wouldn't get to the XP welcome screen), and after several cycles, would finally manage to reach it.

No clear pattern, and of late (not earlier, this was new in the last two weeks), no relation with the CPU or GPU load. Just sitting there showing the desktop, no main application open, could trigger a reboot. This made me mad and angry, and that was when I took it to the shop: if it misbehaves when it's doing nothing, then there is nothing I could do about it. Until I removed the video card...

G

Guillermo, if in your BIOS you have checked the "Auto Reboot", uncheck it.
This way when you'll start again manually your PC you'll have on the screen the hated "Windows Blue Page" that contains many information about the reasons of missed launch.
May be you'll find there the solution to your problem.
Hope this may help.
Bye

Andrea

Guillermo,

I hope I'm not too late for you, but there's a slight risk your computer could be writing junk to random places all over your hard drives before it reboots. I suggest you minimise all activity on your PC until you can eliminate what I write about below as the cause.

I'm having a very similar problem to you and cartrite right now. In my case it seems to be a hardware failure. I get many CRC errors and IO buffer glitches that cause a lot of kernel panics (Linux equivalent of Windows BSoD, but with no auto-rebooting). I found these tended to happen when I moved high volumes of data on any drive.

I've been through the thermal saga to cool the CPU, to no avail. I think the cause for me is a heat damaged / overheating Northbridge and/or Southbridge chip on the motherboard, probably not the CPU itself (mine can fry eggs at >90°C and run stably no problem). Since I'm getting IO errors on my drives (even USB!) I suspect the Southbridge. For you, the graphics card suggests the Northbridge. If you can put a different graphics card into the PCI slot and find you still get reboots - it's very likely the Northbridge.

I suspect many computers die because it's these chips on the motherboards that fail: they need cooling like the CPU, but have nowhere near as much attention to that as them. Once they get too hot, the IO is corrupted.

I have also disconnected all my drives with important data and files (so I can't run Celestia), I do not run memory/CPU intensive applications, and am basically just surfing the simplest web pages (no flash, etc.) to minimise IO traffic.

Even then, I can't be sure. I used memtest86+ and got RAM errors too. These were always on the same locations on the same module - I pulled it out, even though I cannot be sure memtest86+ was fed spurious data because a few data lines got noisy.

A Linux OS will show up such errors much sooner than a Windows OS. If you can get a Ubuntu Live CD/DVD, it'll really tell you. Also, the Ubuntu 8.10 Live CD/DVDs have memtest86+ on them.

For me, that's all going to be until I buy a new computer (Oh... decisions, decisions: which one should I buy!?, which one!!??).

Spiff.

Spiff wrote:For me, that's all going to be until I buy a new computer (Oh... decisions, decisions: which one should I buy!?, which one!!??).

Do it yourself, of course. I have never bought an assembled computer.

Fridger

Of course I will! The current one is assembled. A real new computer would make a fuss for my current drives: I'd have to fit them in, get IDE-SATA adapters, etc.

Oh, I was just mocking the purchasing analysis paralysis that the computer manufacturing industry generates - hey, SATA this, DDR3 that! Every three years! I forgot my <sarcasm> tags .

Anyway, let's see how Guillermo gets on!

Spiff.

Thanks, Spiff. I'm most sure that my data are safe, even in the case of disk corruption. I always buy custom made computers, with carefully chosen components. Still, they fail.

Today the wife used the machin for several hours, without problems. I think I have the culprit. I will take both computer and video card to the shop, so they can check both separately, just to be sure.

Thanks, everybody!

G

I'm not sure what you menat, Susan.

Regarding my computer, unfortunately the powerdowns resumed shortly after removal of the video card. The technicians concluded that the motherboard was somewhat compromised. On the eve of a trip, not wanting to leave the wife compurter-less, I changed motherboard and proc, slavaging what I could (disks, cabinet, memory) and problem solved.

Now I have de video card to sell (it's probably not guilty). It's and AGP so modern boards do not take it, but someone my need it. I have just returned from Brazil, Perhaps I sell it before leaving for Spain in a couple of weeks.

Guillermo

abramson wrote:I'm not sure what you menat, Susan.

...
Guillermo

Guillermo,

Susan means NOTHING, since "she" is a bot / spammer. Look at "her" hidden Weblink in the post. I just notified Selden as usual, such that the user susanpret88 will be eliminated and banned VERY soon

Fridger

Thanks, Fridger. I didn't check, and since the topic had been left "open" I felt that some closure was needed. Man, I have so little time these days.

G

spammer username banned & post removed. i've left the responses since they're informative.