PC resets while gaming

P

PHeeNIxx

n00b
Joined
Oct 19, 2004
Messages
52
This one is starting to drive me crazy. During gaming (e.g. DOOM) the PC will reset after some random amount of time (5,10,15 min). System is a few year old Haswell build, and has been solid up until recently. I noticed this in a few games, knew my PSU was nearly 10 years old, assumed it was the PSU, and replaced it with a nice new Seasonic yesterday.

PSU swap did not help - so it wasn't the PSU.

Ran memtest86 overnight - RAM is fine.

Ran furmark for 3h today without issue ... video card may be OK but I'm not 100% sure on this yet. I may switch to integrated graphics to see if the issue disappears, but I don't know if this will really be conclusive even if it does.

Disabled my CPU overclock completely and restored bios to default settings - still need to do some extended prime95 testing.

None of my temps under load look scary.

Windows event viewer just gives me 'critical kernel-power 41' error which means the OS doesn't really know why the PC reset.

I'm running out of ideas here! Anybody experience this?

Relevant system specs:
Intel Core i5-4670K Haswell 3.4GHz LGA 1150 84W Quad-Core
ASRock Z87M Extreme4 LGA 1150 Intel Z87 HDMI SATA 6Gb/s USB 3.0 Micro ATX Intel Motherboard
EVGA GTX 980ti
Mushkin Blackline 16GB (2x8GB) DDR3 1600MHz CL10 DIMMs
Seasonic FOCUS Plus Series SSR-850FX 850W 80+ Gold Power Supply
 
Last edited:
check to see if it's a heat issue by loading both the CPU and GPU. if the CPU overheats, it will usually just throttle, but GPU overheating will usually cause hard resets.

it could also be a mainboard fault.
 
^what he said. and check your cmos battery, a dying one is like a gremlin.
 
KazeoHin said:
check to see if it's a heat issue by loading both the CPU and GPU. if the CPU overheats, it will usually just throttle, but GPU overheating will usually cause hard resets.

it could also be a mainboard fault.
Click to expand...

I thought the GPU would also throttle if it got too hot (980ti)? I will definitely try to test more by loading both at the same time.

However, let's say I can't get the GPU to fail alone, and I can't get the CPU to fail alone, but I CAN get a failure by stressing both at the same time. What conclusion should I draw from this? I guess if the impact of the thermals caused one or both to go even higher than when they were stressed alone, it might be a lead. What if the thermals in a combo stress test were similar to the isolated stress tests for each component? Does that mean I should suspect the motherboard?
 
I had a very similar problem in a Haswell build. When it first started it was maybe once every two-three days, and only while gaming or doing "combined' stress tests (both GPU and CPU). At the time, I thought it was just some funny Win10 nonsense (I had just upgraded). It progressively got worse over time, over a course of about 6 months before I finally just bit the bullet and started throwing parts at it to figure it out. Near the end it would go a few days with no issue, then start resetting every 15 minutes, most often while gaming or under any type of load, but would also sometimes reset if just sitting idle.

Just like you - dropped all overclocks (didn't have one, but Asus has that built in "boost" thing). After that the most likely suspect is PSU, dropped in a nice Seasonic as well. That didn't fix it. Monitored temps/voltages very closely, all looked fantastic. Then I swapped GPUs - no go. Swapped RAM, no go. Swapped motherboards, no go. Swapped CPUs, no go. Fresh install on new SSD, no go....

I ended up replacing every single part in the computer - in the literal sense of the word literal. A lot of swapping parts back and forth trying to find the bad component. I still didn't exactly figure out what caused it. Some magic combination of something something. I'm using that i7 4790 original computer today, the only things different from the original crazy build are the case and CPU heatsink. And those two items are being used in the i5 that I built out of all the part swaps that I gave to my son, and he uses daily with no issue.

Best I could figure is maybe something on the motherboard was starting to short to the case, or the heatsink bracket was stressing something out on that motherboard, no idea.

It took me a lot of time and money to eventually get a stable computer there, I wish I had a better answer.
 
this very very likely has nothing to do with it, but a friend of mine with a new ryzen 1700x system was having a similar issue, he thought initially it was his 280x failing, but turned out it was his VRM heatsinks overheating (extremely hot to the touch, on a open air watercooled setup...) he threw a fan up top to blow air over the vrms, and that resolved his issue....

but like I said doubt that really is your issue, but figured was worth the mention
 
my basic list of hw testing
Basic hw testing
metest86 v7.x
PRime95 small fft
Furmark
Crystal disk info & /chkdsk /b
Prime95 blend + furmark

The last part is to confirm power distribution and cooling is ok. games load both cpu and gpu why you need load them together for testing as well
 
Brian_B said:
I had a very similar problem in a Haswell build. When it first started it was maybe once every two-three days, and only while gaming or doing "combined' stress tests (both GPU and CPU). At the time, I thought it was just some funny Win10 nonsense (I had just upgraded). It progressively got worse over time, over a course of about 6 months before I finally just bit the bullet and started throwing parts at it to figure it out. Near the end it would go a few days with no issue, then start resetting every 15 minutes, most often while gaming or under any type of load, but would also sometimes reset if just sitting idle.

Just like you - dropped all overclocks (didn't have one, but Asus has that built in "boost" thing). After that the most likely suspect is PSU, dropped in a nice Seasonic as well. That didn't fix it. Monitored temps/voltages very closely, all looked fantastic. Then I swapped GPUs - no go. Swapped RAM, no go. Swapped motherboards, no go. Swapped CPUs, no go. Fresh install on new SSD, no go....

I ended up replacing every single part in the computer - in the literal sense of the word literal. A lot of swapping parts back and forth trying to find the bad component. I still didn't exactly figure out what caused it. Some magic combination of something something. I'm using that i7 4790 original computer today, the only things different from the original crazy build are the case and CPU heatsink. And those two items are being used in the i5 that I built out of all the part swaps that I gave to my son, and he uses daily with no issue.

Best I could figure is maybe something on the motherboard was starting to short to the case, or the heatsink bracket was stressing something out on that motherboard, no idea.

It took me a lot of time and money to eventually get a stable computer there, I wish I had a better answer.
Click to expand...

This doesn't sound good. I was reading through your post and getting all excited that you were building up to a solution, haha :) I hope I don't have this experience!
 
You may want to list system specs so we have a better understanding of your particular PC.

If you are running a UPS try running your pc plugged directly into the wall. I have seen some of the smaller units hiccup under heavy loads.

Like sven said prime95 blend and furmark at the same time should more accurately reflect heavy gaming loads than testing individual components.

If the system is stable enough when not gaming you may want to try updating the motherboard bios just to be safe.
 
Kardonxt said:
You may want to list system specs so we have a better understanding of your particular PC.

If you are running a UPS try running your pc plugged directly into the wall. I have seen some of the smaller units hiccup under heavy loads.

Like sven said prime95 blend and furmark at the same time should more accurately reflect heavy gaming loads than testing individual components.

If the system is stable enough when not gaming you may want to try updating the motherboard bios just to be safe.
Click to expand...

Added relevant system specs to OP.

I'm not running a UPS, but I am using a 'mature' power bar, so I'll try going into the wall directly. The system does appear to be stable while not gaming, for now. Last night I updated the MB and reset the bios to default state, but it didn't help.
 
Is the new seasonic PSU something like the 520w S12 or better? The 350 and 430 units could theoretically provide enough power but don't leave much margin of error.

Another thing I wouldn't rule out would be general Windows 10 wonkiness. Things like the fall creators update and intel meltdown \ spectre patches have been causing stability issues on some system. Some more basic things like setting power mode to high performance and updating all system drivers could be worth a shot. Just as an example Large Windows 10 updates knock out the audio driver on my laptop and I start getting random system crashes because the system reverted to generic audio drivers. Re-installing the manufacturer drivers resolves the issue every time but it took me a long time to track down.
 
OK, so I got home and did some mixed testing. This led to GPU thermals in 80-83 degree C (which as far as I know is normal for hardcore stress testing of the GPU). From what I've read, once it gets into this temperature range, it should start throttling to prevent damage and overheating, not reboot/crash/die.

In any case, I got to the point where I could reproduce the issue within a min or two, usually with visual artifacting in the furmark window a few seconds before the system reboot, which definitely points to the video card. The last test run I did seemed to have completely toasted the GPU however - if I try to boot with it as the primary video device, it doesn't even spit out a signal any more. If I boot with integrated video as the primary video device, I'm able to get back into windows fine.

I assume this means the video card was in the process of dying a slow death, and I just sped it up by stressing it.

I guess I'll have to put EVGA warranty to the test now.
 
Last edited:
PHeeNIxx said:
I assume this means the video card was in the process of dying a slow death, and I just sped it up by stressing it.

Is there any value in trying the GPU in a different PCI express slot? Could the failure somehow be related to the slot and not the card?
Click to expand...

I don't know that a different slot would tell you a whole lot. If your able to install it in a different computer, that would tell you a good deal though.
 
You must log in or register to reply here.
Back
Top