Help me solve a mystery! Failing display (gpu? psu? mobo?)!

evilneb

Commendable
Nov 30, 2016
3
0
1,510
Yo! I've read a lot of posts so far looking into the issue I've been having. So far, I haven't found anything that hits it right on. Help?!

Build:
Crucial Ballistix Sport 16GB DDR3-1600
Intel Core i7-4790 3.6 GHz 1150
ASUS Z97-AR LGA 1150 Intel ATX Motherboard
Cooler Master Hyper T2 Compact CPU Cooler
Nvidia Geforce GTX 660
512 gb crucial mx100 ssd
Corsair rm650 psu

Built this just a little over one year ago. First PC build coming from laptops, so it's been a dream.
Everything in it was bought new (some reburb/clearance from microcenter) except for the gpu, which was a donation from a friend. (I know 16gb of ram was probably overkill).

I've played a variety of games, from WoW and Overwatch to Far Cry 3 & 4, The Witcher 3, etc. Everything has been great until a few days ago. Part of my initial confusion with this was because it only started after rearranging the deskspace and transporting the rig from a friend's house. Powered the machine up again and noticed my dvi to hdmi cable was maybe damaged a bit and display signal wasn't going through. So, bought a new cable. Everything seemed fine at first, but then after being in game, started getting these green flickering pixel lines across the screen. Tried ignoring at first, but it got worse and worse in minutes until the whole screen would flash green pixels, which seemed attached to the character models in game even. Then the display would go black, still having game sound and ability to use voice chat. If I'm able, I can exit game and it'll stabilize again. Nothing seems to be overheating. GPU maxes between 75 and 80, cpu doesn't go over 50 and is usually lower.

I didn't pay much attention to other sensors in gpu-z before, so I can't really say what looks normal except that previously, my temp would never go over 60 or so. And it seemed like a lot of the gpu clocks and load graphs were very high (load and power consumption bump around 100% a lot), but again, I hadn't really checked them closely before.

Regardless, I figured my gpu must be dying, so I got a Geforce gtx 950.

It seemed like this helped. GPU load seemed a bit lower, so I thought I was out of the woods, and I played for an hour or so in different games without issue....until it happened again. I quickly shut it down right away, afraid something was happening that could fry the new card.

Since then, I've reinstalled the old card and it's been a toss-up for when I get issues, and I feel like I notice slight screen tearing and chunkier graphics that I didn't use to notice, but maybe I'm just paying closer attention. The other night, I played various games off and on for a couple hours, and then tried raiding in WoW and had it happen again after another 30 minutes or so.

I've also been doing a lot of googling and reading forums for people talking about similar issues.
I've checked cables, the cards have been installed and reinstalled, the 12v, 5v, & 3.3v seem almost perfect, temps seem fine (though, again, higher than I was used to). Video drivers have been uninstalled and updated a number of times. I've run virus checks and checked startup programs as much as I know how.

I still feel like the coincidence of moving the pc around a bit, damaging a video cable, and also having hardware fail at the same time is too unlikely, but I can't see how the monitor could be the problem. As I said, I also got a brand new hdmi to dvi cable, and even if I should look at them, I don't see how those could start failing only when the gpu is under load for a while.

As far as I can tell, my next step is just to start buying replacement parts and swapping things out until I find a culprit, but I'd really rather not do that. I don't overclock (at least I've never actively picked an option or altered settings in an effort to overclock).

I took video on my phone of the most recent instance of this, with a look at gpu-z immediately after exiting:
https://drive.google.com/open?id=0B6TCI93PSa0dbjB5N25WMkJjX0U
https://drive.google.com/open?id=0B6TCI93PSa0dZmZlV3cxbkpWQ1U
https://drive.google.com/open?id=0B6TCI93PSa0dRHE5eENUc0J6eWs

Anyone have any ideas? I really hope it's something stupid so I can have a laugh.


 
Solution
I'll have to agree with your initial statement on your PSU; that looks fine. So I think we can safely rule that out as a possible cause.

If memory serves, if the graphics card is peaking at 100% it's because the CPU is demanding more of it. Why the power spikes up above 100% seems odd but I've come across mentions about power limits which may vary depending on the manufacturer. I think if temperatures are within an acceptable range it should still be safe.

Although unlikely, what's the cooling for your case like? You've mentioned the slightly higher than preferable temperature of the graphics card, so it may be worthwhile trying to make sure the air flow is good for your case if you haven't checked before. Still unconvincing as a...
Not much to add. You've tried the things I would have tried, but did you try another PCI-E slot in your attempts? Just in case of the possibility it's the PCI-E slot causing trouble.

The artifacts initially sounded like a loose graphics card, though further description made me think it's a VRAM issue. Though if you've researched it, you should have come across those possibilities already.
 

evilneb

Commendable
Nov 30, 2016
3
0
1,510


I did swap to another slot on the psu when changing gpus at some point. Also, I figured that anything like vram would be solved when I put in the new gpu.

Any chance it could be an issue with the motherboard? Seems like when issues arise, the gpu is hotter, like 80, but that's not supposed to be a failing temp. Maybe there is a combo of issues, where it's not controlling the gpu properly and/or reading temps wrong? I haven't updated drivers for that since i built it....

Also, this might be unrelated, but since I built it, the computer has always had a little quirk on boot where the leds and fans come on for a second, then shut off for a couple seconds, and then auto-boots just fine after that. I've always ignored it since it's never seemed to be a problem. Every once in a while (like weeks or months apart) it randomly crashes and boots into BIOS, but always restarts just fine. I figure that's another thing that just comes with pc life sometimes.
 
That rules out the PCI-E slot being an issue then. Don't have much else to add but bounce ideas about...

I didn't put too much stock into the artifacts because it isn't quite like the VRAM issues I'm aware of. I would agree that if the GTX 660 was a problem, the GTX 950 shouldn't be. Though it seems a lot of threads point towards the graphics card when such issues arise with artifacts. Any chance of testing the cards on another PC (a friend's PC perhaps)?

Going back to your original post... you mentioned how you acquired the components for this build. What about the power supply? Discounted or refurbed? (I think you can anticipate where I'm going with this.)

I would anticipate many forum regulars would suggest you invest in a better power supply because, I think, that particular one is considered a bit below par; specifically tier 3. Incidentally, what are the variation on voltages you are seeing when you made those readings. I think I came across the suggestion 3V and 5V should be in within a 5% difference, and the 12V within 10% (not entirely sure, to be honest).

EDIT: have you by any chance tried running a RAM test like Memtest?
 

evilneb

Commendable
Nov 30, 2016
3
0
1,510
5v: Min 5.0, max 5.04
3.3v: min 3.23, max 3.28
12v: min 11.90, max 12.10

I haven't run a RAM test yet, but I will take the suggestion and try later tonight.

Most recently, I've been using HWMonitor to keep an eye on stuff. I notice that the GPU power spikes up to 108%, and the usage is regularly at 99% depending on the game I'm playing. If the psu is maybe not providing enough power to the gpu, would that happen?
I was under the impression that it's not so abnormal when overclocking to have the gpu power go above 100%. However, as I said before, I've not tried to overclock anything.

 
I'll have to agree with your initial statement on your PSU; that looks fine. So I think we can safely rule that out as a possible cause.

If memory serves, if the graphics card is peaking at 100% it's because the CPU is demanding more of it. Why the power spikes up above 100% seems odd but I've come across mentions about power limits which may vary depending on the manufacturer. I think if temperatures are within an acceptable range it should still be safe.

Although unlikely, what's the cooling for your case like? You've mentioned the slightly higher than preferable temperature of the graphics card, so it may be worthwhile trying to make sure the air flow is good for your case if you haven't checked before. Still unconvincing as a possible cause of those artifacts though. But looking at that clip... it does seem either the graphics card or the monitor connection to it. A loose connection can lead to intermittent display problems.

Do check on RAM just eliminate that possibility when you have the time. At least it will narrow down the possibilities.
 
Solution