Sign in with
Sign up | Sign in
Your question

Having a Pretty Big Problem... PSU or GPU?

Tags:
  • GPUs
  • EVGA
  • Systems
Last response: in Systems
Share
April 16, 2013 10:14:52 PM

Had this issue happen twice so far.

While gaming 2 different times on not very demanding games Torchlight 2 and DotA2, just randomly my computer screen will black screen crash and my GPU fan will spin to what sounds like 100%. The first crash the computer restarted itself after 15ish seconds. The 2nd occurrence, I had to manually restart it.

I've contacted both EVGA and my PSU company and troubleshot the issue...

After the 1st occurrence and talking with a EVGA rep.

1. We've tried clean driver sweeps and installs for my GPU.
2. Clean sweeps and reinstalls of Precision, HWMonitor, Speedfan.
3. Checking my PSU +12v, which is what the EVGA people think is the issue.

In the BIOS, it ranges from 11.69 to 11.93 depending on when I look, it will fluctuate while in BIOS as well. When it says 11.93 that generally stays stable, when it shows lower it will jump around. When I told the EVGA people this they said that this was what they believed to be the issue. Although in an email they sent, they also stated that this range was "acceptable", but the person I talked to said this was actually "unacceptable".

Although they didn't ask for it, this is what my Speedfan "usually" shows me, take into account this isn't under any load other than having a random twitch.tv stream open.




Again this isn't under any sort of load other than twitch.tv stream with basic hardware acceleration under the flash options, basically at most 5% according to Precision. I also know that no software/bios reading will be 100% accurate compared to an MM which I don't currently have to test the PSU.

When it is under load the reading spike even lower, but never above 12. But the spikes are quit dramatic. No clue how to infer much from this, besides putting together a computer and basic troubleshooting, this is a little past what I'm comfortable addressing by myself.

Anyhoo, before contacting my PSU company, I first tried all their troubleshooting. And a day later, randomly I got another black screen crash and GPU fan spinning to 100% that this time required a manual restart. Right before that happened Speedfan showed my +12v dip near 11.4-11.5 under load.

After that I contacted my PSU company, and talked to the guy who RMA'd my first 2 units which were defective. At first he seemed skeptical that PSU could cause a black screen crash making my GPU fan spin to 100%. But when I gave him my +12v readings he agreed that seemed off, and the fluctuation seemed off.

So in conclusion, he's sending me a replacement, this would be #5 including returns to the retail store during first 30 warranty because of spotty fan issues.

Anyhoo my question is... Does this seem like it actually is the problem. Or does this sound like a GPU issue?

I'm skeptical because while no one was certain they seemed to want to defer the problem. The PSU people wanted to say it was the GPU, the GPU people were adamant it was the PSU.

Since the PSU people seemed to think that whatever +12v readings I'm getting can't make the GPU fan spin 100% or whatever.

Here's my specs, everything is stock settings... I built this rig 10months ago and have no issues other than random PSU fan issues, this current PSU has seemed to work fine for 7months.

i5-3570k
EVGA GeForce GTX 670FTW
Crucial 256gb SSD
AsRock Z77 Extreme4
Crucial 2x4gb @1600 RAM
Thermaltake 750w Modular Smart Series 80+B PSU
Windows 7 Home Premium 64bit

Any help would be greatly appreciated :)  If anyone could offer any sort of a reassurance that I'm on the correct path, and/or any information of what actually could be the cause if this isn't. I've pretty much tried everything that anyone has said, granted this has only happened twice, I won't try any game that puts the GPU/PSU under load until the new PSU comes, for fear of well, I don't know but not everything is still under warranty.

Again, thanks for your help.

More about : pretty big problem psu gpu

April 16, 2013 10:33:56 PM

GPU or RAM. I don't think it's a PSU issue.
April 17, 2013 3:10:18 AM

Little bit of a follow up... Had the computer running playing some stupid .swf game since I'm staying away from putting the GPU/PSU under much load.

While running Speedfan during that time roughly 4 hours, I noticed that the voltage was progressively getting lower and lower. First hour or so it would fluctuate between 12.0 and 11.7, next hour 11.9 and 11.6, next hour 11.8 and 11.5, til finally it was fluctuating between 11.7-11.4 at that point I restarted the computer and checked the BIOS which showed me 11.2-11.4 fluctuations.

I then Shut Down it and waited 15 minutes and started it back up and the BIOS then showed me 11.9... Speedfan showing me fluctuations between 11.9-11.8 during the first 30minutes and then it started that trend downwards again.

I'm assuming, and again this is just going off common sense and logic, with no indepth PSU/CPU/GPU knowledge... That maybe when my game blackscreen crashed, it was when it was approaching these voltages and if I hadn't shut down it, eventually it would have happened a 3rd time?

I don't know... It's gonna be 3-4 days before the new PSU comes. Am I risk
Related resources
April 17, 2013 4:47:11 AM

What you are seeing is called Voltage Ripple, which means that the voltage coming from the PSU is fluctuating. This is fairly normal, but according to either ATX or 80+ specification (cant remember which) the positive rails are not meant to deviate more than 5% from their rated voltage (which is 12v in this case).
Technically this means that your PSU is doing what it should, it doesnt dip below 11.4V which is the minimum to meet the criteria.
http://www.hardwaresecrets.com/article/Everything-You-N...
However as the Thermatke guy said, that is getting very close to becoming unacceptable.

Voltage Ripple wont outright kill your components unless it goes nutz and oscillates violently, but it wont do any favours for the lifespan of your components.
April 17, 2013 5:10:54 AM

manofchalk said:
What you are seeing is called Voltage Ripple, which means that the voltage coming from the PSU is fluctuating. This is fairly normal, but according to either ATX or 80+ specification (cant remember which) the positive rails are not meant to deviate more than 5% from their rated voltage (which is 12v in this case).
Technically this means that your PSU is doing what it should, it doesnt dip below 11.4V which is the minimum to meet the criteria.
http://www.hardwaresecrets.com/article/Everything-You-N...
However as the Thermatke guy said, that is getting very close to becoming unacceptable.

Voltage Ripple wont outright kill your components unless it goes nutz and oscillates violently, but it wont do any favours for the lifespan of your components.


Thanks for the reply. Do you think this is contributing to the 2 crashes I've had?

Or does this sound more like a GPU or Driver issue.

Oddly enough I did do a clean driver install... But since then I've actually gotten 2 normal crashes related to the drivers I'm assuming, still trying to figure out what the hecks going on with that. One after installing the new drivers, opened firefox and it crashed. Another just out of the blue it BSOD'd me. Neither of those black screened, or made my fan go crazy like the original problem.

April 17, 2013 5:23:12 AM

To be honest I dont know, but I can imagine voltages dipping too low which causes the card to run unstable, leading to a crash.

Drivers have nothing to do with Voltage Ripple, thats entirely up to the PSU and it doesn't use software in any capacity I think.

BSOD's are usually a sign of a failing HDD or a corrupt OS, might want to check up on those possibilities. But unless it starts BSOD'ing fairly regularly, you can just put it down to a random occurence.
April 20, 2013 8:46:08 AM

Little update to this thread.

About 3 hours in with the new PSU, haven't had a crash yet, granted I haven't tried a game in that time, but have watched a few streams for an hour.

I did run the OCCT memtest that crashed once, and made it through that, but that's just 1 short test.

Also notice the difference in +12v on this PSU compared to the first one, which is encouraging.

RMA'D PSU



NEW PSU




So... Don't want to jump the gun here, but it's looking "promising". Hopefully I was wrong with it being a GPU issue.

Although I am using the 310.70 drivers now instead of the 314.22. And I'm using the bare minimum of driver installs, just the Graphics Driver and Physx, I didn't install the HDAudio/3D crap/Nvidia Update/etc.

So really hoping my issue is resolved, but not gonna declare it fixed after only a few hours.

I'll update this, if I end up getting another 0x000000116 BSOD... Which then would mean logically the only thing left would be the GPU.

Thanks for everyone's help.
April 20, 2013 9:53:16 AM

Is anything overclocked, changed from the standard settings?
Do you use a CoolnQuiet feature (or similar "green" feature) for the PSU?

A black crash might infer that your monitor cannot handle high FPS, you'd need a higher resolution than 1440 x 900 and a 120mhz monitor.

Not a lot of other things to consider except DUST! which seems to be quite a large issue that most people ignore or feel its not important enough.

You might want to consider adding a few snapshots of the PC innards, just in case cabling is tangled, or that there is something you have overlooked.

If the video card is cramped between other things, that could impede air flow and circulations.

I doubt that the general performance of the PSU or the video card is at fault, more likely cramped conditions inside the PC or dust or too many external devices attached or that you use multiple OSes or that your free space of the primary hdd is less than 25%, or that you use VMWARE or similar and that your ram is insufficient to provide adequate memory for virtual drives.
April 20, 2013 10:04:41 AM

TenPc said:
Is anything overclocked, changed from the standard settings?
Do you use a CoolnQuiet feature (or similar "green" feature) for the PSU?

A black crash might infer that your monitor cannot handle high FPS, you'd need a higher resolution than 1440 x 900 and a 120mhz monitor.

Not a lot of other things to consider except DUST! which seems to be quite a large issue that most people ignore or feel its not important enough.

You might want to consider adding a few snapshots of the PC innards, just in case cabling is tangled, or that there is something you have overlooked.

If the video card is cramped between other things, that could impede air flow and circulations.

I doubt that the general performance of the PSU or the video card is at fault, more likely cramped conditions inside the PC or dust or too many external devices attached or that you use multiple OSes or that your free space of the primary hdd is less than 25%, or that you use VMWARE or similar and that your ram is insufficient to provide adequate memory for virtual drives.


Everything's stock.

The crash was a 0x000000116 BSOD Kernel 41 which is "display" related.

Further research and others have stated the "general" causes are...
  1. Usual causes: Video driver, overheating, bad video card, BIOS, Power to card


Drivers were clean installed, reinstalled, clean installed to older version.

Overheating not even related as, CPU temps during the crashes were 30-38c, GPU temps were 30-45c.

Rig was just built 10months ago and is cleaned regularly, cable management and air flow is clean.

As to being a black screen, it was only black screen because apparently the crash being "display adapter" related, it shut my monitor connection down.

EVGA was adamant it was the PSU not supplying proper voltages, and so far it's looking to be the case.

Although a few and myself included are still skeptical that maybe the GPU is still a possible cause... I won't know for sure unless I get another crash, but I'm hoping this new PSU fixes it.

Still haven't tested this in any sort of game or putting the GPU under some sort of load... Probably will try that later tonight.
April 20, 2013 10:20:01 AM

Sooo, what were you doing when the crash occurred, how many apps were opened, internet and how many tabs etc.. media player?

Do you switch between a PC game and another app like internet or something or is the PC game the only thing activated?
!