Pc shutdown while gaming

[invisions]

hey folks,

i've had a problem with my pc for sometime now, while playing games (bf3 etc.) my pc just crashes after 5/10 minutes (i have to turn the psu on and off before i can boot again).
recently i formatted the harddrive and performed a clean install (32->64) of everthing again (the problem existed before this aswell, but the increaseUserVa for 32bit solved it).
i had no trouble running assassins creed 2(and the expansions) hours on end, but with battlefield 3 the pc just crashes aswell as with world of tanks (which isn't that demanding of a game).
i've been searching on forums and blogs to find an answer/solution to my problem but i haven't found a conclusive one.
answers range from a dieing psu to a overheating gpu/cpu.

specs:

win 7 64 bit
abit ax78+ - coming up on 4 years old
amd phenom x4 2.5 - coming up on 4 years old
evga gtx285 - 2/2,5 years
samsung 320 hd - coming up on 4 years
6gb ddr2 800 - recently added 4gigs
750 watt ms-tech psu - i had an other psu before this (it shortcircuited) 2,5 years old
2x dvd-roms - 4 years old
4x 120mm casefan(no-led) 1x 140mm casefan

note i have upgraded the standard cpu cooler to an scythe mugen II

now as for the results of several tests i've run on the pc:
furmark with hwmonitor - (gpu tester + pc monitor), the results were:

voltages never change more then 5% (not a single one)
power consumption on the cpu stays the same
the only real suprise was that the read outs on the motherboard stay stable but have sudden "spikes" (stable ~50C then a 120C spike)
while the cpu cores are 45C max
the gpu however that thing cooks up to 87C when stressed (fan @91%)

note the gtx285 is HUGE! it is a tight fit in my casing and it is larger the the mobo itself in (26cm orso) and the fan sits right over an heatsink on the mobo


i hope anyone could help me fix my computer or give me advice on what to do next.
if there is any more information required please say so.


regards,


steffan

 

[Sunius]

Sounds like a memory issue. I'd try checking it first. You can download the memtest tool from here:
www.memtest.org

Make sure to let it pass to 100% at least 3 times!

 

[simon12]

Again Sunius beat me to it thats just what I was going to suggest. If it passes that try running OCCT, Intel burn test or Prime 95 and check your CPU max temp and if it crashes. If thats all OK look at control panel>Admin tools>computer management>event viewer and see if there is any indication there, or you could check this forst as its quicker than all the above.

 

[invisions]

thanks for the quick response, i've been trying to get the memtest to work but so far no succes.
both on usb and cd it won't boot and thus i cannot test it, really starting to rage about that.
instead i have now begun testing on prime 95, i've also taken out the oldest 2gig stick to see if it made a difference, i didn't.
and i've another remark to make, the more i underclock the gpu the longer it plays without crashing.

will continue testing and post the results once i get them.

 

[invisions]

i've run prime 95 for 6 hours now, no results, no changes in voltage over 5%.
hwmonitor still registers 140C as max temp for CPU while i've never seen it go over 70C (cores never go over 55C).

as mentioned in my first post my previous psu shortcircuited, could it be that my mobo got a shock aswell and now the consequences are showing?
when i check the event viewer what should i be looking for exactly?
any other advice?

 

[Sunius]

How did you create a bootable flash drive/CD disc? How are you trying to boot it?

 

[invisions]

cannot make bootable usb, need linux to do so.
i burned it on cd, but it won't boot

 

[simon12]

Have you set yur bios to boot from CD?

 

[Sunius]

And you didn't just drop the ISO file to the CD, did you? You need to right click it and press Burn Disc Image (on the .iso file!).

 

[invisions]

i've got it up and running now

 

[invisions]

ran the test 5 times, it passes 5 times.
what now?

 

[MajinCry]

I had the same problem with my old cpu/mobo. The problem was that under CPU Ucore Voltage, it was on auto. Usually, that's okay. But for my processor, it was being under-volted, so the computer shut down when I did something like Crysis (But it played Skyrim for hours just fine). I looked up my processor's spec sheet (Mine was a core 2 duo e6700 3.2GHz Wolfdale), found the voltage specs, and entered that in my BIOS.

Give it a shot, mebe?

 

[Sunius]

Next in the line is Power supply. Do you have a friend with good enough power supply to try in your PC?

 

[invisions]

i have done as majincry said and increased the voltage of both my cpu and ram. haven't crashed since. haven't had time to put it to the max to be sure yet though.
i'll do that during this week just to be sure.

 

[invisions]

ok it didn't solve the issue although it has greatly increased the time before it crashes

 

[Sunius]

Could you show us a screenshot of HWMonitor?

 

[invisions]

here is the screenshot:



Uploaded with ImageShack.us

or

http://img96.imageshack.us/img96/7/hwmonitorx.jpg

 

[Sunius]

By how much you increased the RAM voltage man? That is serious overkill, revert the RAM back. Afaik DDR2 default is 1.8V?

Did you test your power supply in another PC? By any chance you've got a multimeter?

 

[invisions]

changed ddr2 back to normal, the time before it crashes is significantly shorter while with the 2v it crashes after a longer time if it crashes at all.

i did not test the psu from another pc since i have no way of getting the required wattage from anyone i know.

i could ask a friend of mine with a multimeter, what would you like me to test?

 

[Sunius]

With 0.2V overvolt you can easily fry the chips.

And if you can get a hold of a multimeter, test voltages of a molex connector when the PC is on. Yellow should be 11.6-12.4V while the red should be 4.8-5.2V. Black ones are ground.

 

[invisions]

reset the ram, checked the psu.
i started tweaking a bit in the bios and the pc runs fine now, although i still can't run the gpu normally (still underclocked).

 

[cats_Paw]

My first bet would automaticly be the PSU, since that is about 70% of the stability problems i have ever heard about.

Another thing that might be causeing this is the GPU having a corupted ram module, but i dont know if you can actually test for that.

On anotehr note, your cpu cores only go up 3 C from idle to stress? never seen anything like that, i think you might be doing something wrong.

 

[invisions]

the screen was taken in a brief moment while gaming (bf3) because after 5/10 minutes
it would crash, now i can make you another one after a couple hours of gaming.

btw
i have 5 casefans, blowing at full capacity and a scythe mugen II cooling the cpu directly.