Random freezes, random auto-restarts

[Ferruruccruio]

The title basically explains what's going on.  For months, ranging from every few weeks to, sometimes, intervals of a few days, my computer randomly restarts itself, or freezes completely.  Any idea what is causing this?  My RAM is perfectly stable, according to memtest86 and prime95.  I've gotten no bluescreens whatsoever.  This problem is so random, without any pattern on what causes it, that I have no leads on how to reproduce it.

The only idea I have is the power supply, but I don't see how the power supply would cause freezes.  I do see how it would cause restarts, though.

[SW Chris]

Memtest doesn't test GPU RAM, afaik.  Is there a utility that does this?  That could be the culprit.

You can test your power supply with a voltimeter, see if any of the leads are wildly out of the 10% "safe zone" or are fluctuating.

Also look for any capacitors on your motherboard that may be bulging.  The black cylinders with the silvmetal tops.  If one is, that's your problem. 

[Dr. Tax, Ph.D]

Do you have 'Auto restart upon bluescreen' selected in the system Startup and Recovery settings?

[Commie]

How random we talking about? Does it reboot when it's idle and your afk? Does it seem to happen during light use (web surfing, etc) or only under load (gaming)?

[Ferruruccruio]

How random we talking about? Does it reboot when it's idle and your afk? Does it seem to happen during light use (web surfing, etc) or only under load (gaming)?

It happens no matter what the use is.  I've come back to my computer seeing that none of my applications are open anymore, indicative of an unannounced restart.  When I AM using the computer, it crashes under both heavy and light load.

Do you have 'Auto restart upon bluescreen' selected in the system Startup and Recovery settings?

I do not have auto-restart enabled.

[Talisan]

1. Install and Run Ccleaner (www.piriform.com/)
   a) on the Cleaner Tab, check everything except Wipe Free Space, then Run Cleaner
   b) Click the Tools Tab and uninstall any software you no longer use
   c) Click on the Registry Tab, check mark everything(or do them one at a time so the list isn't too huge)>Scan for Issues, then Fix Selected Issues.
   d) On the Tools Tab, click on Startup, then disable any startup programs you don't think you need. Don't delete them, just in case.
2. Goto My Computer, Right Click one of your drives then Properties, then click the Disk Cleanup button
3. Uninstall and Reinstall your video drivers(might even do 1c again inbetween the uninstall and reinstall)
4. If you have a sound card, consider uninstalling and reinstalling those drivers.
5. Defrag the drive. You can use the Defrag program that came with Windows, or download one from (www.piriform.com/) called Defraggler which works great for me.

You said your memory was rock solid, but if it was not then I would consider re-seating. Same for the video card.
If you have an air compressor, I would consider blowing out the inside of the case to remove dust. You can use canned air, but the hazard there is sometimes it comes out as a liquid.
I have experienced bad power supplies before where a computer will reboot unexpectedly. Swapping it with a spare and seeing if there is a difference can determine that. You could do what Chris suggested and use a volt meter to check if the power is within specs, assuming you had a volt meter, knew the specs for your supply and had the experience to do it.
I have had a video card go bad as a result of a capacitor that popped, as Chris suggested. Replacing the video card solved the issue. Replacing the capacitor would require finding one of the same type, getting a solder gun, and having a really delicate hand. Prolonged heat can ruin other components near the points you will need to unsolder and resolder.

[ʗαртдɨɳ ṂДÐ ɱїкє]

Canned air only comes out as a liquid if you do it wrong I've never had it accidentally do that on my hardware, and even then at one place that I worked, we tested our components by spraying the supercooled propellant at them. I don't think it really affects the hardware as much as it would your skin.

[SW Chris]

Even capacitors that are just bulging (and haven't exploded) can cause random instability issues.

[Ferruruccruio]

1. Install and Run Ccleaner (www.piriform.com/)
   a) on the Cleaner Tab, check everything except Wipe Free Space, then Run Cleaner
   b) Click the Tools Tab and uninstall any software you no longer use
   c) Click on the Registry Tab, check mark everything(or do them one at a time so the list isn't too huge)>Scan for Issues, then Fix Selected Issues.
   d) On the Tools Tab, click on Startup, then disable any startup programs you don't think you need. Don't delete them, just in case.
2. Goto My Computer, Right Click one of your drives then Properties, then click the Disk Cleanup button
3. Uninstall and Reinstall your video drivers(might even do 1c again inbetween the uninstall and reinstall)
4. If you have a sound card, consider uninstalling and reinstalling those drivers.
5. Defrag the drive. You can use the Defrag program that came with Windows, or download one from (www.piriform.com/) called Defraggler which works great for me.

You said your memory was rock solid, but if it was not then I would consider re-seating. Same for the video card.
If you have an air compressor, I would consider blowing out the inside of the case to remove dust. You can use canned air, but the hazard there is sometimes it comes out as a liquid.
I have experienced bad power supplies before where a computer will reboot unexpectedly. Swapping it with a spare and seeing if there is a difference can determine that. You could do what Chris suggested and use a volt meter to check if the power is within specs, assuming you had a volt meter, knew the specs for your supply and had the experience to do it.
I have had a video card go bad as a result of a capacitor that popped, as Chris suggested. Replacing the video card solved the issue. Replacing the capacitor would require finding one of the same type, getting a solder gun, and having a really delicate hand. Prolonged heat can ruin other components near the points you will need to unsolder and resolder.

1: What led you to believe that these symptoms are indicative of a broken registry?

2: How would not having enough free space on my disk cause random restarts or freezes?  It found thumbnails and temporary internet file stuff.  I cleaned it, but I am 100% sure that it won't lead to freezes or restarts just having them on my disk.

3: Already did, didn't work.

4: Don't have one.

5: Windows 7 does its own auto defrag.  I'm not fragmented at all, right now.


RAM doesn't need reseating, as I indeed said that prime95 and memtest86 showed no issues.  Reseated graphics card very recently, problems still persist.

Already replaced the graphics card.

The only thing I haven't done is replace the power supply.  I would need to get another one that's around 850 watts.  If that's the most likely culprit, then I can just go through with it and get one.

All my capacitors look fine, from what I can tell.  It'd be hard to tell on my graphics card, though, without taking the big shroud off, and that's protected by "void-warranty" tape.  The warranty has already gotten me one free replacement graphics card.

[Talisan]

Step 1 is what I automatically do whenever I have problems and was suggested to me by others way back when. Not a bad idea to keep the registry clean. Also, uninstalling unused programs can remove a bad driver that could potentially be causing a random boot. Doing a registry clean also removes most of the keys left over after uninstallations that leave crap behind. I do Step 1 at least once a month and that usually ends up removing a quarter to a half gig of junk. Also, a hard drive becoming full can cause all sorts of random problems with Windows XP and older, especially if you had notifications turned off and didn't know it. I know you said you are using Win7, so I'm just talking.
 
Most of the time when my PC's have gone bad over the years its a hard drive going bad. I run my computers 24/7 for prolonged periods of time so they are constantly under load. The only time a bad hard drive ever caused a reboot for me was when Windows drivers or files got corrupted. Again, this doesn't apply to you.
 
Based on everything you've said, my bet is on the power supply. When I bought a brand new case from MWAVE years ago, it came with a brand new generic power supply. After I built the computer and during the Windows installation I was having all sorts of rebooting issues. I went down to Bestbuy and bought a new PSU and came home and replaced the generic unit. No problems after that at all. I did not get my money back for the generic PSU as it would have cost me more to ship it back than it was worth. What I recall being odd at the time was the generic PSU weighed very little in comparison to the Thermaltake PSU I replaced it with, which to me indicated a definite quality difference. Both units were 650W, and I am still using that computer today.

[ʗαртдɨɳ ṂДÐ ɱїкє]

I do agree...I think most software failures will do other things before your system randomly reboots. If it's bad enough to crash the Windows kernel, you'll bugcheck. I think its definitely a hardware issue if the system is rebooting without any software failures.

[Ferruruccruio]

Kinda funny, since the power supply was the ONE THING where I was told that I was paying too much attention to.  It was emphasized to avoid the generic brands and go for some highly rated and respected brands, and other than that, just pick one and forget about it.  I picked the Seasonic M12D 850w power supply because it was highly rated in a number of fields, like efficiency and noise.  It wasn't cheap, and I figured I was getting a solid brand.  Maybe not??  I might just go with Corsair, now.

[Commie]

I'd be curious to know if you had that problem outside of a Windows environment.
You say it reboots even when your afk and it'd idle?

You could perhaps try to just leave the computer on sitting in the BIOS overnight, and see if it reboots at some point. If it does then its definitely a hardware issue.

It does sort of sound like a power issue, but that doesn't always mean its the PSU's fault, and your current PSU sounds like its pretty good already.

[Talisan]

Seasonic M12D 850w you say? That has modular cabling correct? I've heard of people having problems with units that allow cables to be removed. While I can't remember what article I read, I do recall that there was a complaint with some loose connections at the PSU end. I know from experience that the more connection points or splices you have in a wire, the higher the failure chance. A soldered connection vs a pressure based connection is far superior.
 
Modular cabling seems like a fantastic idea because you can remove any unused cabling from the case and improve air flow, as well as making the insides of the computer look nicer and less cluttered.
 
Lastly, I've never heard of Seasonic before, not that that means anything, but a few years ago I did do a lot of research when buying power supplies for some computers I was building for some friends friends, and I never came across that brand. I'll have to look into them some more and see what people say.
 
Edit: After reading a few articles on the brand and that specific unit, it looks like it's a great buy with a high efficiency rating, etc.

[Commie]

I have an Enermax MODU 82 now but my next PSU will be a Seasonic. The Modu use to be the most efficient PSU on the market, but Seasonic holds the throne now.

[XgF]

I do agree...I think most software failures will do other things before your system randomly reboots. If it's bad enough to crash the Windows kernel, you'll bugcheck. I think its definitely a hardware issue if the system is rebooting without any software failures.

Not necessarily. The x86 CPU has several exception levels:

• Normal fault. These are things like page faults - they happen in applications all the time (virtual memory and such) and aren't unusual. One in the kernel/a driver is bad and will panic/bugcheck most systems
• Double fault. Something is really screwed with one of your normal handlers - i.e, a fault happened while raising a fault. These are unrecoverable (panic/bugcheck guaranteed)
• Triple fault. Something is really, really screwed up - not even the double fault exception handler could be raised, everything has gone wrong, and the system is coming down now. These cause the processor to reset the system, since there is no way of handling them.
  

If its a bad driver issue, the latter could be the culprit.

[Ferruruccruio]

Not necessarily. The x86 CPU has several exception levels:

• Normal fault. These are things like page faults - they happen in applications all the time (virtual memory and such) and aren't unusual. One in the kernel/a driver is bad and will panic/bugcheck most systems
• Double fault. Something is really screwed with one of your normal handlers - i.e, a fault happened while raising a fault. These are unrecoverable (panic/bugcheck guaranteed)
• Triple fault. Something is really, really screwed up - not even the double fault exception handler could be raised, everything has gone wrong, and the system is coming down now. These cause the processor to reset the system, since there is no way of handling them.
  

If its a bad driver issue, the latter could be the culprit.

Well, ok, but that's not really helpful!  Thanks anyway!


I have another lead.  I just had another auto restart, but I was doing something *relatively* unusual this time.  I was running both AW and SL, along with firefox and chrome, so, basically, under a lot of "load".

If it's the graphics card, that will royally suck, as this is already a replacement graphics card.  I don't even know how I'd nail it down as the definite culprit.  If I COULD convince the warranty department that the card is at fault, then I'll just switch the card manufacturers and sell this XFX card.

[SW Chris]

Brainstorm idea:  Keep those four programs open whenever you're at the machine so you can recreate the circumstances as best as possible.  If it happens a few more times under those conditions and it seems to be happening more frequently, I think you could say it is your graphics card (or the PCIe slot on the mobo, perhaps?). 

[ʗαртдɨɳ ṂДÐ ɱїкє]

If its a bad driver issue, the latter could be the culprit.

I actually wish Ferr was bluescreening. I've gotten pretty good at using WinDBG from my work. At Microsoft. On the team that fixes Windows

I know we make the driver development kit and require driver signing specifically to reduce the amount of utterly crappy drivers out there, and I would find it hard to believe that a genuinely-signed driver would be capable of screwing up so badly that the system can't even produce a BSoD. I mean, we're talking about a serious, low-level failure that's preventing the kernel from writing directly to the storage medium, in space it's already holding and knows how to interact with. I'll have to ask about this at work next week...

Either way, you could always try to run a Linux CD and see if you have the same problems. Only difference is that your storage medium is your optical drive instead of your hard drive, but asides that the OS is running on the same hardware setup, so that should be able to place blame on the PSU or not.

[Ferruruccruio]

Had another auto-restart.  This time, after logging back into windows, and starting up MSN, the monitor instantly shut off, and a display driver failure was reported.

I don't know if these two events are related.

[SW Chris]

I should have thought of this much earlier, but I don't suppose you have "Automatically restart after a blue-screen" enabled?  Disable that so you can actually see when the error happens.

Go to My Computer, Properties.  Click on Advanced System Settings.  Under Startup and Recovery type Settings.  Uncheck Automatically restart.  Click OK and OK again and close System Properties.  Next time you have an auto-reboot you should actually come back to a blue screen and you'll be able to find out what the offending driver, file, or error code is.

[Ferruruccruio]

I should have thought of this much earlier, but I don't suppose you have "Automatically restart after a blue-screen" enabled?  Disable that so you can actually see when the error happens.

Go to My Computer, Properties.  Click on Advanced System Settings.  Under Startup and Recovery type Settings.  Uncheck Automatically restart.  Click OK and OK again and close System Properties.  Next time you have an auto-reboot you should actually come back to a blue screen and you'll be able to find out what the offending driver, file, or error code is.

I said, in a previous post, that this was disabled.

[Commie]

Disable it, Ferr. 

[SW Chris]

I wasn't sure. lol.  So it's not blue-screening.  That makes it difficult.  And there are no error messages in the event viewer system log? 

Have you tested your power supply and all of its connections with a voltimeter or power supply tester?
Do you have your system plugged into a UPS?  The UPS could be faulty.
No nicks or dents in your IDE or SATA cabling?
Hard drive checks out fine with Seatools?
Memory checks out fine with Memtest after 3 or more passes?

Can you get a cheap video card and run your system with that for a week or so?

[ʗαртдɨɳ ṂДÐ ɱїкє]

Any luck, Ferr? I brought this up with a coworker and the opinion is you either have bad hardware, or possibly bad firmware on something. Either way, either your PSU is bad and causing the machine to reboot, or there's something that's failing so badly that as Xgf pointed out, you're reaching the "triple fault" stage, and your system is trying to bluescreen but failing at doing so.

My mentor @ work used to work with PCI hardware and says that its possible to make the machine triple fault if the CPU is executing code stored on the hardware itself (firmware?), but that it's very rare to see this in Win7 like you're seeing.

Either way, if this is still happening to you, I would run with as minimal a configuration as you can get and try to take note of what the system was doing when these reboots happen. Even if it's idling, there are always processes doing things in the background like AV scans or Windows indexing. Also give the linux CD idea a try if you haven't already. That could help rule out an OS problem.