PC crashing/hot GPU [Solved]

KurnDerak

Getting specs out of the way:
MSI R9 290x 4GB
16 GB Ram
i7-4790K 4 core 4GHz
Corsair CS550M PSU
Dual monitor 1-3840x2160 @30Hertz 2-1680x1050 @60Hertz
win10 64bit

Boring more info: CPU tends to stay 35-38C on average across the cores, idle or brief time in game. Motherboard is the same. GPU idles at 68C, jumps to ~95C by the time I get to the menu in something like Borderlands: Pre-Sequel. I got it down to about 60C before by forcing fans to stay at 100% and slightly turning down memory clock and such. Did furmark GPU test, no crash. 95C almost instantly. Did several tests from 10 seconds to just over a minute. Did a HeavyLoad CPU test. 1:25, CPU maxed at 68-72C across the cores. Dropped back to 40 seemingly instantly upon completion. Also no crash. Have done a memory diagnostic on startup, no issues there. Checked event viewer. Only thing I've found is something telling me the previous shut down wasn't expected.

I had previously been able to play more system intensive games such as GTA5. It would blow out noticeably warm, almost hot air but with something like Skyrim it barely reaches what I'd call warm.

Each crash is exactly the same. It's like the system instantly shuts off. No lights, no fans, no beeps, nothing to monitors, keyboard and mouse lights go off.

I didn't see any problems until a few days ago after I started playing Skyrim w/ mods. I thought it was mods at first because of the timing until I crashed in another game. Outside of crashing everything appears to run just fine, no FPS drops, no graphic problems, no warnings, etc., etc. I've had some Skyrim CTDs but with no warning with those I expect those are mod related. Figure I should mention them though because of the timing. The amount of time in game until I crash varies wildly. When I first tested Borderlands: PS it crashed before I could even get past the main menu. Other times I've been able to play Skyrim for 4-6 hours before I eat a shut down.

Right now I'm trying to eliminate all options that I can before I end up spending any money. My current thought is that the thermal and/or fan setup for the GPU is failing (no idea for how long or how much it contributes to crashing) and the PSU is now failing the system's power needs during gaming. If that were true, would I need to change up the fans on the GPU or would changing the thermal paste be enough?

wunderbar

95C is a "normal" temperature for that GPU under load. I know it seems hot, but GPU's run hotter than CPUs, and that particular generation was hot and power hungry. I double checked some reviews for the card and they agree that 95C is not at al out of the ordinary for the card, especailly on a reference cooler.

The first thing to try is re-installing the graphics drivers. You could have a corrupt driver that is causing your issues. After that it could be PSU, but my gut is telling me it's GPU related. Possibly some bad DRAM on the video card, so when the game you're playing tries to load assets into the bad RAM chip on the card it crashes out, which could explain why the time to crash varies.

Again, try the graphics drivers first. That's a relatively easy, and more importantly free test. If you have another video card, even an older one that can't run the games at great settings, I'd toss that in for a bit as well and try to see if you can break it.

Foomy

Those symptoms don't sound like a gpu crash or gpu overheating to me.

What it sounds like is that your psu is dying and as it heats up it's hitting a ceiling where it can't provide enough power and it hard crashes.

tsmvengy

Yep, sounds like a PSU issue to me. If it were GPU I would expect tearing and graphics errors before it shut off.

KurnDerak

My plan right now is to replace the PSU, tentatively looking at Thermaltake Smart 650W since I can pick it up locally, and soon replacing the thermal paste on the GPU. My worry about the GPU wasn't really that it was reaching 95C but more how quickly it was. It would think if it was something wrong with the RAM on the GPU it wouldn't just shut the whole system down every single time, or that it could happen with something like streaming video or such, which it hasn't thus far. Correct me if I'm wrong on that assumption. Somewhat same with something being wrong with the drivers. I would also expect something in event log or some kind of error if it was software based outside of it telling me something in the event viewer to the effect of "the previous shutdown was unexpected".

Thanks for the replies.

Foomy

gpus heat up quick, especially if your running stock fan profiles as they tend to favor keeping the noise down over lower temps. So I wouldn't worry too much about redoing the thermal paste of your gpu unless you really want to do it anyway.

And ya if it was a gpu crash you would normally see either tearing, artifacts, other weird graphical glitches, or windows will report that your gpu driver has stopped responding.

KurnDerak

Replaced the PSU, things seem to be running just fine now. Idle temps appear about the same (CPU seems a little higher right now, but don't know if its something variable I'm not seeing causing it to process more or what), however in game GPU temp is a little bit lower. In Skyrim it's maxing at 91C, staying mostly in the mid 80s. Still going to replace the thermal on the GPU, it's cheap and makes me feel better. CPU increases only a little during game. So it appears changing the PSU thus far has solved the problem. Unless I get more crashing or something I'm going to call it good for now. Thanks everyone.