My computer turns off randomly (boring technical difficulty)

category: general [glöplog]
Hey all, I was hoping you'd help me diagnose a strange problem I've been having.

Since I replaced my Athlon 64 x2 3800+ with a 64 x2 5600+ that I got cheap, my machine has started turning off whenever it got loaded, and some times just randomly. For example, running the Win 7 Experience index benchmark crashes it around the DX10 Geometry Shading test.

I decided I must have a bad powersupply with spikes, so I calculated my needs and went out and got myself a Corsair 400 watt PSU. It arrived today, but it did not solve the problem!

I removed a couple of RAM sticks, but that didn't do the trick either. Checking my CPU temperaturs after a crash, they are quite low (40 degree celcius or so).

Any ideas? I have a 8800GTS 512 MB Alpha Dog and 2 GB Corsair RAM now (NVidia demobox :)), and my BIOS is the latest.

Could be nvidia punishing you for getting a free demobox and not releasing at nvscene in the end :D
added on the 2010-06-09 23:40:41 by keops keops
I once experienced that on a crappy laptop where the heat wasn't being transported properly away from the CPU, causing it to twist/deform a little and losing the connection on a pin or two - this caused the laptop to turn off. I actually had to shake it in order to reconnect the CPU properly :D

(Compaq, wohey!)
added on the 2010-06-09 23:42:45 by leb00ster leb00ster

@bstrr: Thanks, I'll check for that tomorrow, maybe try my old stock cooler over my fancy Scythe thingy :)
Gaga: glad I could help, it's probably another issue then :)
added on the 2010-06-09 23:52:58 by keops keops
Have you checked the temp for the GPU?
added on the 2010-06-10 01:04:08 by Deus Deus
clean out the fans / heatsinks on your gfx card.
added on the 2010-06-10 01:23:27 by spiny spiny
Get a air-in-can and go over your rig meticously.
added on the 2010-06-10 02:33:53 by Deus Deus
i had a similar problem once - i fixed it by a) replacing the psu with something bigger and b) making sure the vents where clean
400W isn't exactly a lot of power these days, if the other things doesn't work out, i suggest testing your system with a bigger PSU (600w+) just to be sure. 400W should be enough, but its worth a try.
added on the 2010-06-10 06:58:28 by dwarf dwarf
65W is enough for everyone.

added on the 2010-06-10 08:48:25 by leb00ster leb00ster
This problem stuck me many times...

The first time, was a loose RAM (erm... I hit the side of the computer, after pressing power, that was below DELETE, when trying to delete a polygon while modeling, without saving for 6 hours... this loosened the RAM...)

The second, was a RAM that my sister put her filthy fingers on... (she do stuff on impulse... she wanted to test more ram on her machine, went to mine, removed it carelessy, and dropped the 2000volts that human body store on my precious RAM contacts).

The third time, was a cooler not properly placed... the fourth, was on the same machine, the person that fixed it, don't put new thermal stuff on it ¬¬ I had to do it myself in the end.

The fifth time, I have no idea of what was going on with the machine, I still don't have, this machine still exists, and also it destroyed 2 motherboards, 1 processor, 2 PSUs, 1 nVidia card, 1 HD (not at the same time... in fact, the machine is not even the same anymore, only the case is, and it still randomly reboots and destroy parts... it is even creepy to think about it...)

The sixth time, happened to a machine, that now I am using as router (that machine although great, don't supported SATA 2, and I only noticed that after buying a nice 500GB SATA 2 disc... So I got borrowed a new machine, I am typing from this new machine). This router machine, still randomly freezes, shut-downs, or refuse to boot (it may freeze on bios screen, or don't even turn-on the screen...). And it had this problem since it was new... I installed more coolers, changed the PSU, checked the CPU, switched motherboard, RAM... probably the CPU is faulty (a Pentium D).

I still plan to figure a solu
added on the 2010-06-10 08:59:29 by speeder speeder
speeder: sounds like you're cursed.
added on the 2010-06-10 09:24:08 by v3nom v3nom
I'd say cooler not properly installed or PSU not large/stable enough. Try checking the voltages while running a benchmark. If the voltages drop below what they should be your power supply is too weak.
Also you should do a BIOS update to make sure the board runs the CPU with all the proper voltages and shit.
added on the 2010-06-10 09:30:00 by raer raer
Do you know for a fact that the CPU was fully functional in another system? Have you updated your motherboard BIOS (first thing I would try) to the newest avaible version? Did you do a clean install of the OS upon switching?
added on the 2010-06-10 14:25:33 by NoahR NoahR
sorry, Im asleep.
added on the 2010-06-10 14:28:53 by NoahR NoahR
digital narcolepsy!
I have to say that I have never sucessfully installed a new CPU without eventually having to reinstall the OS because of too many quirks related to drivers (I suspect). And rather than spending hours upon hours hunting down each indivdual problem, I just spend 1 hour reinstalling and usually that fixes my problems.

Does it always reboot when doing graphics related tasks? You mention a DX10 test. Did you reinstall Nvidia drivers?
added on the 2010-06-10 14:51:27 by NoahR NoahR
download rightmark cpu whatever util and check cpu heat. If the machine turns off it's an overheating problem. You must keed your rig cool.
added on the 2010-06-10 15:01:17 by chriz74 chriz74
Hey all, thanks for all the replies.

@eeblis: I just noticed that the DirectX 10 benchmark *always* crashes at the exact same point. I monitored my temperatures and voltages using Speedfan, and they seemed alright (~43 CPU, ~60 GPU). Hence I suspect that it's a software related issue, and not a hardware related one. But I actually never checked the CPU on another system...

I'll try re-installing my GPU drivers. If that does not work, I'll re-install Win7, seems like the only sane thing to do.

Again, thanks :)
had the same problems.very often. on my home pc, a 700w psu and lowering the dram-frequecy did the job .
tweaking the bios to non standard cpu-fsb-ram-voltage settings did the job very often too.
mostly it was some weak ram who could not handle it.
try to underclock your system and test it again. then go step by step back to the standart settings.

despite that, on a normal system the 8800GTS Alpha Dog needs at least a 500W PSU to work well. now go and get a new PSU

Sounds similar to when my GPU-fan was broken.
added on the 2010-06-10 16:30:09 by r0XX0r r0XX0r
Seems like eeblis solved it: GPU driver re-install did the trick. Buying a new PSU was pointless (but then again, my PC noticeably more silent now :)). Thanks all. Hope it stays like this, or it's a full re-install :P
Or not... Still crashes after longer periods of usage (gaming though!), and my south bridge is very very hot. I also just noticed that it just so happens to be right under my GPU and passively heated. Recipe for disaster, anyone?
So could it be that my new CPU makes the southbridge run at a higher frequency, and that it thereby makes it operate outside its limits, causing it to overheat?