Computer just powers off under high graphic load

[themaxxz]

I hope you can give me some insight in a problem I'm having.

The problem is that my PC turns itself off, when doing graphic load intensive things. So far I experienced the problem while playing
- team fortress 2 using high gfx settings
- running ntune stress test (one time)
- running 3dmark06 stress test (one time)

I do not believe it's heat related to either CPU or GPU as the poweroffs seem to also occur while the GPU and CPU or in a normal temp range. CPU at 38degrees, GPU at 68degrees (captured using the ntune log feature)

Specification of the computer (newly built in aug 07)
- Thermaltake Armor JR. Zwart, Aluminium Front, 2x120mm Fan
- Gainward GeForce 8800GTS 640MB GDDR3, PCI-Express, "BP8800GTS-640M-TV-DD"
- Corsair TWIN2X 6400C4 DDR2, 2048MB CL4
- Thermaltake Toughpower 750W, Modular
- Asus MB P5N32-E SLI PLUS I775A nF65
- Intel CPU Core2Duo 2,33 BOX E6550 4M
- HDD 320GB 16MB Baracuda Sata ST3320620AS Seagate
- running everything at stock values (not overclocking CPU or GPU)

Steps already taken:
- manually configured memory timings as specified on corsair site for said memory 4-4-4-4-12-22-2T at 2.1V
- also ran with auto memory timings
- dropped memory to 667Mhz (1T)
- updated to the latest directx
- updated windows xp sp2
- re-installed windows xp sp2
- installed latests nvidia drivers
- installed mem DIMMS in slot1 and slot2 to run in single channel mode
- performed memtest86+ which finds no errors after 2 passes
- ran orthos prime overnight and found no errors
- ran atitool for to stress my video card up to 80 degrees celcius and didnt crash
- ran Lost Planet Extreme Condition Trial for close to an hour and found no errors
- upgraded to the latest BIOS
- changed pci-e express connector on PSU per suggestion on thermaltake forum


Could this be a PSU problem even though it's new and rated for 750W, way more then what my PC actually needs? Or do you think the MB or CPU may be bad?

Unfortunately I have no other PC where I can interchange parts with to rule out specific components.

 

[seemingly random]

Impressive and thorough amount of upfront work. The psu is way overpowered but have never heard of this causing problems - just waste.

It seems unlikely that the mb or cpu are bad since so many things in 'steps already taken' work. Is there any one test that can consistently cause the pc to shutdown? The one thing not listed is removing the mb and all components from the case to eliminate any possible shorts and trying to reproduce the problem - probably a stretch but it appears that all other bases have been covered.

Other than this, you might try an exorcist...

 

[robisbell]

what are the temps for the CPU, Video Card, and Motherboard?

you can use http://www.cpuid.com/pcwizard.php to check them, and if you be so kind as to post them.

 

[themaxxz]

The temps reported by ntune when my PC shut off were,
CPU 38deg C
MB 30deg C
GPU 67deg C

 

[themaxxz]

Oh, and the one thing which always seems to crash my PC in a short time (under 20min) is team fortress 2 with in-game resolution set to 1280x1024. Initially I believed it was just and issue with this game, but then I managed to reproduce the problem with ntune stress test (1 time out of 10 attempts or so) and with 3DMark06 benchmark run (1 time out of 6 attempts or so).

 

[robisbell]

your video card running at 152 Fahrenheit, extremely too hot, it's cooking itself to death.

 

[themaxxz]

That was under load, idle time is around 55degress Celcius. From reading other forums this is an acceptable temperature for nvidia 8800GTS cards. It's even run hotter up to 80degrees Celcius without any ill effects.

 

[robisbell]

I've spoken to 2 tier 2 techs at Nvidia, and they said 152 will damage the card. they said it's rated for a high temp but for short term only, for gameplay it should not stay that hot for long periods. I see non amd forums a few people don't care but I also see quite a few that are concerned and have had graphic issues and all seem to be heat related, I think I saw one where they person pulled the card and the heat had warped it.

 

[themaxxz]

I've done as suggested and removed everything from the case onto a table. With 2fans 12cm blowing and this setup and by underclocking the 8800GTS to 500Mhz, the system hasnt crashed yet. Below are the temperatures during the gameplay.

Current Time CPUºC) SystemºC) GPU1ºC)
2007/11/30 3:8:50 27 29 46

Now I just need to get all this back in the case and add some additional cooling I guess.

 

[robisbell]

well, somehow the temps became frowny faces, but I am going to still say it's a heat issue.

 

[seemingly random]

Originally posted by: themaxxz
I've done as suggested and removed everything from the case onto a table. With 2fans 12cm blowing and this setup and by underclocking the 8800GTS to 500Mhz, the system hasnt crashed yet. Below are the temperatures during the gameplay.

[reformatted for lightweight bbs s/w]
CPU: 27c
System: 29c
GPU1: 46c

Now I just need to get all this back in the case and add some additional cooling I guess.

Those temps look great. What is the ambient temp?
What are the temps if no case fans are blowing on the mb - there should be one blowing on the hd though.

Before you put everything back into the case:

I would set the clocks of the cpu/gpu exactly as they will be in the case - and retry everything.
How long has it been tested? long enough to be quite sure that it's heat related? Has everything been tried that caused it to fail in the case?

I'm not convinced yet that it's heat related. The temps from the 11/26/2007 10:18 AM post weren't out of line. I would try to artificially cause temps to increase to verify this. Try setting ASUS q-fan to silent in the bios for the cpu if not there already. It looks like the stock hsf is being used.

Also, try a copy of speedfan - it has the capability to keep a history of temps, etc.