random crashes

[tom.mullie]

My computer has been having random crashes for a while now. I'll be doing something (doesn't matter what), then I'll get the BSOD, it will "dump physical memory to disk" (the wording may be a bit different, it only lasts a second) then shut off, restart, get as far as loading windows and then die again. It will loop until I flick the power/unplug and restart. Then it will work again for an unpredictable amount of time.

Its the OS that kills it (x64 Vista, although I don't think it matters) since it never crashes until windows tries to load, but I think the cause is hardware, since there is no predictable trigger, that is to say no specific program or action that will kill it.

If I let it cycle enough times it will sometimes (but not always) bring up a BIOS message saying that my overclock failed. The only think OCd is the RAM, which is only brought up as far as the factory specs (it defaults to slower, but same voltage). From having a simmilar issue on my laptop a few years back, I'd guess MOBO, but I can't back that up.

I've tried running benchmarks of various subsystems to try and provoke a crash, but nothing happens.

After all that, any thoughts? any and all suggestions appreciated.

[Exmortis]

Memory, constant random crashes I always suspect memory, do you have access to some tam you can borrow to test with?

[tom.mullie]

I don't have any to borrow. I'll look into finding some, but in the meantime I've bumped up the voltage a bit (just 0.05V above the recommended). That was only half an hour ago, though, so we'll have to wait and see.

[Chilly]

Do a favor for us, check the windows logs, you can get to them by "Start -> Control Panel -> Administrative Tools -> Event Viewer". It should look something like this:



Let us know what kind of errors(if any) the event log has recorded. The answer may be hidden in those logs. That being said I'm leaning towards a possible defective ram theory.

[enaberif]

Wait. your running your memory at spec'd voltage then bumped it up? Shouldn't have to do that.

Have you ran memtest or p95 to see if you have any system instability?

I see your running a P35 board and Crucial memory; you may perhaps have crucial memory that had a problem with a P35 chipset.

But I would for sure run memtest or even p95 and see if you get errors.

[Exmortis]

Originally Posted by enaberif

Wait. your running your memory at spec'd voltage then bumped it up? Shouldn't have to do that.

Have you ran memtest or p95 to see if you have any system instability?

I see your running a P35 board and Crucial memory; you may perhaps have crucial memory that had a problem with a P35 chipset.

But I would for sure run memtest or even p95 and see if you get errors.

Actually he may have too, Thanks to JEDEC's inability to write clean specs, most memory requires over voltage these days. Hell my old 4x1GB required over voltage of .7 just to boot and run. My DFI board over voltages my memory by a small margin automagicially because I have 4 sticks. I have heard others need a small voltage rise even with 2. JEDEC isn;t that great at writing electrical specifications, there was a great White Paper on this issue on Corsair's site.

[tzetsin]

I'll second Enaberif's suggestions, and add removing the vid driver with driversweeper, and reinstalling/updating it. I suspect the ram, but i've seen bad vid driver installs cause random crashes too, especially if you tend to leave your computer on all the time. When you look at that log file, if it has anything that says "nv" in it, its definatly the vid driver.

[tom.mullie]

Yeah, bumping the voltage did nothing (I tried bringing it up another notch. no effect at all). I haven't run memtest, only the Orthos memory test. It didn't die for about 10 minutes (20min total uptime is about par for the course right now, so it didn't shorten it at all). I'll try to get a hold of memtest and see.

as for the error logs (last 24h): 81 errors
the causes of errors seem to vary- no one operation is mostly responsible.
a list of the repeating ones:
WMI: 20 times
Httpevent: 22 times
Service Control manager eventlog provider: 18 times
eventlog: 15 times
other than that, a few random applications did it once or twice

I don't think its a board compatibility issue. Asus specifically lists this RAM as compatible, and I built this thing back in April, with no troubles until late October. I don't know what the voltage was before I started playing with it (ASUS boards only tell you that its "auto" by default, not what it automatically chose.) I've got it at 2.1V right now, and Crucial recommends 2.0. Doesn't seem to have made a difference. I'll get memtest and post results soon. Thanks guys

[tom.mullie]

well, memtest ran rock solid for 40 minutes, 0 errors. As for the video drivers, I updated last night, after the troubles started, and nothing changed.

so the memory's probably alright. any other suggestions?

[sswilson]

Heh 40 minutes doesn't guarantee that the memory is good, but it's a start at least.

I know folks hate to do this, but have you tried to re-install the OS? If it's a driver issue, you could spend days/weeks trying to track down the problem while an OS install shouldn't take more than 4 hours or so.

Not everybody will agree with this method of faultfinding, but it would certainly help to point towards hardware if the problems continue after a fresh install.