Computer "crashing" when games launched

Operating Systems & Open Source Windows
1

So, heads up this is apparently a pretty complicated issue and I've already tried about everything I can think of for trouble shooting on my own, as well as going to another imageboard for help.

A while back, my computer would crash if I launched afterburner, no big deal I thought. It's overclocking, these are beta drivers, whatever. I let it slide to focus on finals and I dont really need my overclock outside of 3 games. The issue then became the computer crashing whenever I launched a fps or GPU monitoring program at the same time as a game (steam in game fps did not cause crashes, however)
I figured this was a driver issue, so I uninstalled with DDU, and installed the stable drivers, same problem
So I began more in depth measures, 3 different drivers, on 2 different bioses, each uninstalled with DDU and reinstalled correctly, completely uninstalling afterburner, cam, fraps, everything and ccleaner'ing after. None of that worked
Recently I thought it might have to do with my system OC, so I brought everything back to stock, same problems. Today I reinstalled windows and started from scratch. Without the OC on the CPU, things work... sometimes. Some games still cause a "crash"
This "crash" isn't really a crash, but rather, the game freezes and gets stuck in fullscreen, although everything in the background still works. If I attempt to close the game in task manager, the computer locks up and i have to do a hard reboot, the system never really "crashes" per se.
The issue is, there shouldn't be any issue with the OC, under IBT im stable for 4 runs on very high and 30 min of max (16gb of ram) and the system is stable in furmark, which stresses the GPU far past any game. So something is happening either in drivers or somewhere else
Watching the power draw in furmark I draw up to 750w, in IBT i draw up to 400, but power delivery should be fine, given my PSU
Memtest86 showed no errors for one pass, will be running for a full 8 hours when I go to work tomorrow. This tells me the issue is either the mobo, CPU, or GPU. I'm inclined to think it isn't the mobo, since the bios update didn't change anything.

At this point I'm virtually out of ideas in terms of what the hell the problem is, I'll be glad to give you guys more info as long as I have it or can get it. I really just want to be able to play my games and work on my computer without worrying about all this again. Is there someway for me to get an error report? or test if my CPU is dying?
Will be doing smart tests for HDD/SSD health tonight, and I can get a multi meter to test the PSU tomorrow too

tl;dr - my pc will freeze games, and when I try to close them the whole thing locks up. Bios updates, driver updates and rollbacks, removing overclocks, and reinstalling windows have all failed to fix the issue. GPU is stable in furmark, CPU is stable in IBT overclocked or not, but together not stable in games.

My PC
8350
Asus Crosshair V formula-z
4x4gb g.skill sniper 1866
295x2
WD black 1tb - media games etc
Crucial mx100 512gb ssd - boot
seasonic 1250w x series gold
that's about all that's relevant at this point, I'll post cooling if you guys need it.
I can also share the info on my OC if that seems like it could be important

2

What I can think of right now are two things:

1) Use a less powerful GPU (preferrably something without any external power needed) and use it to test the PCI slots;
2) Try to connect the 295x2 to the other 16x slot on your motherboard and see if you get any crash.

I don't think you're having issues with your power supply but what I would do is give a try with another one but I know would be a pain in the ass to take everything off.
Also test your drive's health is a good idea.
Another thing: I don't know if you could do that but maybe try to "disable" crossfire and see what happens.

1 Like
3

Tested the drives, they're completely fine, no problems. I was just thinking of disabling crossfire, I'll give that a go today. Also, I hadn't thought of reseating, I'll try that too. The problem with testing another psu is I'm not sure I know anyone with a psu strong enough to run my system, I forget if my friend has a 850 or 750w psu, so I'll have to ask. I was thinking that I might put my brothers 290 in and give it a whirl to see if I cant rule out the gpu
At this point I'm pretty sure its either the motherboard or gpu, just a matter of ruling out the gpu.
my buddy also just got a 950, so I'll see if I can borrow it to test the pci slot's if I dont make any progress otherwise

also, general update, I tried flipping the bios switch on the gpu, it did nothing, as to be expected.

4

Another update. After seeing that the computer is fine in either furmark or IBT, I decided to run both. Bone of the things I noticed just running furmark on its own was that my gpus were downclocking, so I went into afterburner and raised the power limit, not voltage, to 150%. This allowed it to keep its clocks. Running just furmark the pc draws around 1000 from the wall on my kill-a-watt. putting the cpu into the mix with ibt on very high, the power draw increased to over 1100, but then rapidly falls and stabalizes between 650-750, with the gpus still maintaining their clockspeeds, but the video performance grinds to less than a third of what it was. If I stop the benchmarks, then restart the same thing happens
I think this rapid power drop might be part of the issue, but I dont know. Do you guys have any thoughts?

5

look into event logs

6

I would check the event logs first for clues. If nothing turns up there, I would check the pins in all the power connectors for corrosion, discoloration or signs of overheating. On a rig that is drawing that much power, the connectors can often become a weak link in the chain. Even bad connectors in your wall outlet can cause trouble, but be cautious of high voltages if you decide to check those...

Clean off any pins in the connectors that don't look shiny and bend the female ends inside the Molex connectors to a more closed position to ensure that they hug the pins tightly. No joke, I often find those blackened pins are caused by loosened female Molex connectors and not the other way around... Anyway, people often forget to check one of the simplest parts in their system when it becomes unstable for seemingly no reason. I've found that connectors, whether they be for power or data, whether going from your powersupply to your motherboard or between the circuit board of a hard drive and its motor etc., are always a main point of failure and cleaning them up and bending them back into place can even bring some dead hard drives back to life.

7

How can I go about checking event logs? what should I be looking for?

8

https://teksyndicate.com/videos/windows-event-log-explained

1 Like
9

look for a error around the time of the crash

10

ok. I'll look at the logs asap. I'm out of town for the weekend but monday I'll try to recreate the events and check the logs. Thanks for all the help so far guys, will report back soon.

11

If you damage anything while overclocking, you can get some pretty weird behavior from the hardware.

12

ok, looking at the logs now I cant seem to find any issues with the cpu motherboard or video card. The only warnings I'm seeing are for my ethernet adapters, and the only errors I get are when I shut the computer down after it locks up, along the lines of shutdown was unexpected. I'm going to try taking all of my expansion cards except the gpu out and try the gpu in another lane, and then if that doesn't work try reseating the cpu

@eidolonFIRE I know OC'ing can damage hardware, the thing is that in any application but games, the computer performs well, with expected behavior, so I don't believe it's damaged

13

just an update, my screen is flickering periodically while web browsing, much like when there is a driver issue or an unstable gpu clock. I'll try reinstalling earlier drivers to see if that solves it, if not I'll go ahead with the re-seating and moving pci-e cards. Before that though anyone have any ideas what might be causing that? or if it's related?

14

if it's still overclocked it could be a symptom of that, if not then you might have a flaky GPU, it happens, I had a 7870 that was flaky from probably day one, crapped out once a week or fortnight at most and only during some heavy work. when you send a card away for warranty they won't test it for long, you want a nice reproducible fault.

15

Wow that's a ton of power draw, first thing I did was look at the PSU but you seem to have a plenty good PSU for the build. I was wondering, you say the system spikes when you use your kill-a-watt to determine the power being pulled from the wall. How many other devices and or other appliances (AC, TVs or other things) do you have going into that same circuit?

Have you tried taking the entire system to the kitchen or somewhere else in the house where you know there isn't alot of load and trying to test there? I know it might seem silly but I used to live in a tiny apartment (NYC) and saw first hand how overloaded circuits would not necessarily go down or trip the breaker all the time but sometimes spike up and down without it being the PC's fault at all. Fluctuations of power might lead to all kinds of abnormalities in a system too, maybe worth a shot to test?

Not claiming this to be the answer, but it can't hurt to try. If this doesn't help I would really try to test with another power supply everything points at it being power related for sure.

16

I've tried unplugging everything including 2 of the three monitors on my breaker, but I want to try it in my brother's room as well, he has newer wiring and is on a better breaker. I did think of that though, good to point out

17

This is definitely sounds like a weird hardware issue.... What is the performance tab on task manager showing when you load up a game??

18

Sorry for the delays, been really busy traveling and with work. I moved the computer, re-seated the gpu, and removed the other pci-e cards (ethernet and tv tuner) and now everything seems to be working fine. Will put the cards back in and see if those were the issue or if the power was, but it seems the issue is fixed!