Possible GPU Overheating, fix?

Jun 23, 2018
5
0
10
I’ve got a Shuttle XPC SZ87R6 case (basically a oven) with an Intel i7 @ 3.5 GHz and a GTX 970. I replaced the back fan recently and added a “fan curve” on my card using Precision.

The problem: after a bit of gaming (games like PUBG, CS:GO, The Division), the card will get hot. I’ve recorded temps up to about 70 degrees without the fan curve and it usually stays around 50/60 degrees with the fan curve (idling around 30). Then, I will eventually get the STOP error code 0x000124 (BSOD). The computer will usually reboot on its own and start up again.

I have blown out my PC numerous times, there is little to no dust. The 970 is about a year old, bought it new. This setup should be working fine.

Any ideas? Is there a way I can fix this problem, or is it simply the design of the case that is overheating the card? I have the original fan that came with the case just laying around (it works fine), should I attach it to the back of the case? I’m open to anything here.

EDIT: PSU 80 Plus 500W power supply
 
Solution
80+ 500w isn't a good psu. I'm going to assume that it's a Evga 500-W1, worse than the Evga 500B. All those 500 80+ are older, group regulated designs, pushing a supposed 480w on the 12v rail. If it's not the Evga W1, then there's a good chance it's not even putting out the full possible 500w.
I7-3770K? Capable of @150w at heavy loads, the gtx970 can run @220w, mobo/drives/fans add another 100w. Pushed to its limits, your pc could see @470w. On a cheap psu, anything close can mean quirks as you run into thermal problems, degradation of caps, low end heatsinks, cheap components with relatively short lifespans etc.

There's a reason gaming designed psus cost more, the 500-W1 being more for Office use, long usage at @200w, not gaming...
Jun 23, 2018
5
0
10
Nice video- however I am still running Windows 7. I have run most of those programs in the past- things like memtest or scannow, all coming back fine. However, I think I’ll try them again.
 
It's some sort of hardware error/machine check exception. Hardware could actually be failing, but if your power supply is marginal then this could look like the same thing. Is your power supply sufficient (and higher temps imply more power consumption)?

Some suggestions on the web tend to suggest overclockers see this, but then overclocking is a fight of damaging the part with heat and keeping the voltage high enough for stability versus a race with cooling (having enough voltage under load to be stable makes it harder for cooling to win the race).

Sometimes such a thing will occur if there is a power line brown-out which a human might not see (if you get an UPS get one with brownout protection). If it happens a lot then it seems more like something internal to the PC.
 
Jun 23, 2018
5
0
10
I think my PSU is not the problem. It should be enough for my rig, as I’ve researched it before, but I can’t remember what it supports- I will post the PSU I have and how much it supports asap. I have always thought it could be hardware. The computer is connected to a power strip, if that’s what you’re talking about? I’m not too sure on that last part of your post.

I have looked at this thread:
http://www.tomshardware.com/forum/68255-63-bsod-issues-assistance-please

Does anyone think this may work? I am nervous to change this due to my PSU, but like I said, I will look at it asap.
 

pyeloyev

Reputable
Jun 24, 2018
13
0
4,520
Not sure if it's the temp from that reading. Mine gets to 105 degrees C and still won't BSOD. They usually throttle themselves when they get hot. Something else could be the problem or maybe you're not getting an accurate reading (like if the sensor is on the heat sink and there is bad contact between that and the component). BSOD can be caused by other components, like memory. I'm afraid you have no choice but to test and eliminate components one by one. Try pulling a memory stick, then the other one.

If you really think it's the GPU, you can try getting a cheap used one like this:
https://www.ebay.com/itm/EVGA-GeForce-GTX-260-896MB-Dual-DVI-Tested-et/292595482721
and playing that game on lower settings and see if it keeps happening.
 
Power supply specs go out the window if failing. Under heat most power supplies will go marginal. Your supply may not be of sufficient quality. I know how many people think that because it turns on and puts out the voltage it must not be failed...you'd be very surprised at how many power supplies fail for incredibly tiny times a human can't even see, and it brings the system down. If you have another supply to swap out as a test it is advisable.

Power strips prevent damage from surges. They do nothing against droops in power...for this you need an UPS with brownout protection. This of course only matters if power in your area varies...often it is tied to locations with air conditioners all kicking in during a heat wave (they are inductive loads).

Quality of power delivery is questionable until you've tried another supply.

 

Karadjgne

Titan
Ambassador
80+ 500w isn't a good psu. I'm going to assume that it's a Evga 500-W1, worse than the Evga 500B. All those 500 80+ are older, group regulated designs, pushing a supposed 480w on the 12v rail. If it's not the Evga W1, then there's a good chance it's not even putting out the full possible 500w.
I7-3770K? Capable of @150w at heavy loads, the gtx970 can run @220w, mobo/drives/fans add another 100w. Pushed to its limits, your pc could see @470w. On a cheap psu, anything close can mean quirks as you run into thermal problems, degradation of caps, low end heatsinks, cheap components with relatively short lifespans etc.

There's a reason gaming designed psus cost more, the 500-W1 being more for Office use, long usage at @200w, not gaming usage where the gpu alone can demand instantaneous high power, then drops.

I run a i7-3770k @4.6GHz with a gtx970 @124% OC, in a define R5 case that's sits in a cubby. Worst possible thermal properties there is for a pc barring sitting it on top of a heater. Powered by a Evga 550w G2 and doesn't even break a sweat.

Seriously thinking the psu is the issue, with any of those ultra-budget 500/600w psus, it usually is.

 
Solution
Jun 23, 2018
5
0
10


The PSU is exactly the same as it was when I got the case, found on the Specs tab:
http://global.shuttle.com/news/productsDetail?productId=1723

Is this the PSU you are referring to?
If so, or if it is just another really bad one, do you recommend I purchase the same as yours? Or another?
Thank you very much for the reply, it does put a lot in perspective.
Also, that is the Intel i7 I have in my machine.
 

Karadjgne

Titan
Ambassador
Eww (sorry), the stock psu is pretty much maybe at the same level as the Evga 500 W1, probably worse. In order to sell those cases with included psus, corners must be cut somewhere. When you figure a decent 500-550w psu is @$50+, a i7-4770 is @$300, Z87 mobo is $100, you are looking at a $450 retail investment. Not including the case. Price tag on newegg was @$350. In order for newegg to make any money at all, you'd be getting low-end Z87 mobo and basically a $10 psu.

Yes, I'd replace the psu with something better, but that's me, firm believer in the single most important part of the pc, the psu. It's the one component that should never be skimped on. Cut corners anywhere else, but never there.