I just built a new computer and installed Windows 8.1. Everything is fine, except every once in a while Windows will just lock up, freeze for about 30 secs or so, then reboot. It happens when I go to switch applications (once just by hitting the Windows button), and when there is a lot of disk activity (downloading from the net or Steam, or installing a program). I don't remember it every rebooting without a download/installation going on (but I am still setting it up, so there is a lot of downloading/installing going on). If I stay within an application, I haven't had any problems yet - and that includes running benchmarks like Star Swarm.

I thought it sounded like it might have been a memory issue, so I ran memtest86, but every test passed.

Anybody have any ideas what to check?

Thanks!

Go into the BIOS, and select the slowest available memorry timings. And don't overclock anything (memory, cpu, bus, etc.).

See if that improves anything.

Good idea. I'll try that when I get home this afternoon.

Because of your description of when and how the bug occurs, I believe it to be a bug in a driver. Perhaps one of the following:
- Video driver
- Network driver
- Hard disk controller driver

A crash like the one you describe should be accompanied by a dump file. Run SYSDM.CPL, navigate to the ADVANCED tab, and hit the settings button for Startup and Recovery. Make sure it's configured to write a dump file ("Write debugging information: Automatic memory dump").

If it's already configured correctly there, then, look for the file C:\WINDOWS\MEMORY.DMP. The file date/time should correspond exactly with the date/time of the most recent crash.

Open up the DMP file in WINDBG if you happen to have that installed. If you don't have WINDBG installed, then I'm sure someone here on the BBS would be happy to analyze the DMP file for you. After the file is loaded, type:

!analyze -v

And it'll give you a verbose analysis of the crash. Among that information is a call stack. Frequently in situations like this you can easily see the name of the offending module right there in the call stack and sometimes a hint as to what the module was trying to do at the time of the crash. The offending module might have a cryptic-looking name like ser2pl64.sys or something, and if you don't recognize it, google for it. (In my example, the name is a USB-to-serial driver for a USB chipset used by some GPS units.)

You guys and your wealth of information never cease to amaze me.

I might have to feign sickness to go home early and try it out.

Also, if you don't feel like installing WINDBG, it seems you can have your DMP files analyzed on the web. Interesting.

http://www.osronline.com/page.cfm?name=analyze

Of course, there is an inherent security risk with giving someone else a dump of your system memory. For instance, your system memory might contain unencrypted passwords and other personal information. But if you think you can trust the dudes, go for it.

It was configured to do an automatic memory dump, but no dump files exist.

I'm going to change the dump file to a mechanical HDD to see if that has something to do with it (as opposed to the SSD system drive).

It may be an un-catchable failure. Sometimes I've had computers that crashed due to a driver bug, but which didn't write dump files because the crash was so bad that it hosed the kernel's ability to write dumps.

If it's really a problem with your SSD failing, then you might never get a dump at all.

I don't know if SSDs have an equivalent of "SMART" but if they do, look at that.

The SSDs came back as healthy.

I did get a crash when I had a mechanical drive selected for the dump, it just hung there for over an hour without producing the dump file (usually it reboots within a couple mins of freezing).

I might just nuke it from orbit and try Win 7. When I tried that last time, the keyboards quit working. The other USB devices worked fine, but any keyboard that I plugged in didn't get recognized, even though they did before the last set of updates. The keyboards still worked fine in the BIOS though.

Non-stop fun

Are you sure the main board itself (or some attached component) is not defective?

I can't be sure, which is why I am looking for tips/help to narrow it down to see what it really is.

I want to say Win 10 worked fine, just the interface irritated me (seriously, who thought that not having black available as a desktop color was a good idea) and Win 7 worked fine until the last batch of updates. That makes me think it is a Win 8.1 driver somewhere that is conflicting.

This might not be it but the wife received a Window "upgrade" in April that caused her computer to crash after being up a few minutes. I put her laptop back to a pre-April state and it works fine.

I'll look at reverting and trying to limit which updates are installed.

My guess is something's wonky with your PSU. I've known the strangest things to happen then. And it doesn't matter if the PSU is brand new or not. If you can, try and swap it with one that is old and trustworthy, see if that changes anything.

If it's a desktop PC, then yes, there might be something to the PSU thing. My desktop PC had a similar problem which went away when I put in a different PSU. It's been a while so I'd forgotten about it.

I had similar problems on one older desktop PC. I agree with Tony that it looks like drivers and I spent quite some time trying different / older / newer ones over time.

It turned out to be a bad DDR module, though.

What does Dance Dance Revolution have to do with it?

Seriously, though: you're right, it seems like it could be a bad RAM problem too. He said he ran memtest86 so I assumed ram was ok. But he didn't say for how long he let memtest86 loop. Sometimes intermittent ram problems only show up after a bunch of loops.

I let memtest86 run for quite a while. It did just shy of 6 complete tests, all of which passed with no errors.

It did just crash right after it woke up from sleep mode when I tried to select the new 'Start' button (first action after waking up), but yesterday I was able to run Wasteland 2 for hours on end without an issue. Basically, it is baffling.

I'd try to run with one module only fora while. Assuming you have two installed.

I am pretty sure I found the problem. Either the new hard drive is bad (I bought a new HDD and SSD for this machine and transferred over one SSD and two HDDs from the old machine) or the SATA 2 port is bad.

I was able to reliably lock the machine up just by waking it up from sleep mode and selecting the Start button. I noticed that while transferring data from the new hard drive, it wasn't waking up after sleeping (drive not responding errors). Simply disconnecting it seems to have solved the problem, and I've been running for 3 days without a lockup.

Now to figure out if it was the port or the drive...

Thanks for all the help, folks. You never cease to impress the hell out of me and this is another thread to bookmark for future reference.

I had a machine with a flaky onboard SATA connector once, caused no end of "unrelated" issues until it was discovered. I touched up the solder and then hot melt glued it to keep it from flexing. It's been good for years now.

Sorry to bump this old thread. I finally got around to fixing the computer last weekend (lack of free time really sucks) and it did turn out to be a bum PSU.

Nice!

Interesting: the few PSU issues I experienced directly all caused a "random" crash and reboot, not the 30 seconds freeze.

Out of curiosity, what PSU where you using?

The symptoms kind of morphed over time. Eventually it got to the point where it would just lock up, but not reboot.

The old was a Cooler Master V1000. The new one is a Corsair RM1000.

To be fair, I also swapped out all the power cables, just in case. So there is a chance that one of the cables was the issue and not the power supply itself.

So the computer I purpose built for the Ashes of the Singularity Beta is finally good to go, a week before release, and right as class started back up so my free time disappeared again

I should have jumped on the bad PSU bandwagon. I had a computer years ago that exhibited a bizarre slate of behaviors, and I couldn't figure out what was going on. In my case it wasn't so much a bad PSU, but when I removed the ATX cable from the motherboard I saw that three or four pins had scorch marks on them!

That was when I replaced everything and invested in a backup battery with good surge protection