PC freezes after long stretches of gaming

[KinesticBat]

AMD 8350 FX
RX 480 Nitro+ 8GB OC
ASRock 970A G/3.1
Corsair CX 750W
Corsair Vengeance 1600 16GB

The CPU I've had for abut 4 years and the PSU for about 3, GPU and motherboard are new.

I play games like Hitman and Fallout 4 and they seem to freeze the entire PC after long stretches of playing, usually after 5+ hours, though with Fallout I can go much longer until a freeze may occur. There are no restarts, just a sudden freeze. Other games I play like Dark Souls 3 or Sleeping Dogs, FFXIV, don't. I might think this is a heat issue were not my temperatures well from the overheat points. GPU during gaming is between 70-75C. CPU is between 45-50C. I can leave the computer on for days doing everything except for gaming and it is just fine.

Prior to this, the computer would bluescreen because of certain games and I had fixed that with an audio driver, now it freezes after long gaming session.

-Memory has passed 10 passes
-GPU(or PSU?) I noticed has a case of coil whine
-Both hard drives have been scanned and nothing bad found
-I've reseated everything, all wires, and changed the wire to the GPU


If this were a PSU load issue I'd expect it to restart, not to just freeze?
And were this a motherboard issue, I'd expect it to crash in any environment, not just gaming?
If this were a GPU issue, why would it choose to crash after long periods?

I am a bit lost of where to start testing and how, and if this even a hardware or software issue now. 🙁

Edit: -I've noticed the inside of my PSU is extremely dusty o .o and compress won't blow it out

 

[RaDiKaL_]

A PSU issue won't always cause the pc to restart, and given you have the CX model it's very likely that's your problem, if you do a quick search in the forums you'll see how many people has had issues with this PSU.

However lets not jump into conclusions yet, time to do some stress tests.

Use IBT-AVX to stress the CPU, monitor temps, cpu values and voltages with HWInfo. Run it for at least one hour.

Use Furmark to stress the GPU, HWInfo for monitoring.

Use 3DMark or Unigine Heaven to stress gpu + cpu thus pushing the PSU, again, HWInfo for monitoring.

Lets see which of these tests causes your system to crash. Enable the log feature so we can review later what exactly happened.

 

[KinesticBat]

I did Furmark and Heaven and 3DMark and had well results and had normal load temperatures.
With HWInfo I had seen something before that had made me worry and I saw it recently again - some of the ' VIN ' , specifically VIN 4 and VIN 6 would go to 0.00V. Is this a sign? I don't think voltage should at anytime go to 0.00V even if there is no utilization unless of course shut down. Maybe I would see this phenomenon(both in testing and gaming) maybe once every few hours. Maybe I think the psu est putting power elsewhere, or it is a sign of a powerloss?

Unfortunately no crashes or freezes. I suspect this would need much longer testing.

As seeing the state of my PSU, it's very dusty inside, and my lack of cleaning it over the years has lead to a nice layer of dust-grime that refuses to be blown away. To both cross it of the list and get a better PSU as you say it is infamous I see lol - I got the <EVGA SuperNova 750W G2>. This will suffice I hope and will be receiving it midweek next week, because Amazon Prime sucks now :|

Oh, also - I saw that my motherboard uses 1600DDR3 OC, and CPU-Z says the ram is running at 1.5V. I know memory is a fickle thing, maybe it needs more? Maybe I'm grasping and should focus on the PSU for now lol.

 

[RaDiKaL_]

Most DDR3 RAM modules run at 1.5v so that seems to be ok.

I don't have much knowledge regarding the VIN values, did you notice any major fluctuation at the vcore value? if its quite a gap (either in the CPU vcore or GPU vcore) that'd be a sign of a problem in the PSU. Same goes for the +3.3v, +5v and specially the +12v values.

 

[KinesticBat]

Hm I don't quite know what is major, but at 100% load of all core, the VCore would plumment from its idle of ~1.4V and go down to ~1.2V and the voltage would dance in that range, never going lower or higher than 1.2V. It seemed to have a pattern to go from 1.26 - 1.28, and then back down, and maybe back up to 1.27 and then back down to 1.26. If it matters, CPU Power would range from ~53 - 58W and then randomly fall to ~30W or ~40W. I do not see +3.3V, +5V or +12V in my monitor. I see only Vcore, Vccp2, AVCC, 3VCC, VIN4, VIN5, VIN6, 3VSB. I don't know how to translate those, but they are all measured in V.

VIN4 and VIN6 I notice follow each other whenever they change, maybe a short ms in lag, maybe it corresponds to a rail. I also booted up Hitman recently. Both of those values I mentioned fell to 0.00V from 2.016V 30 minutes in and sustainted a 0.00V until the computer froze. It's never happened so quickly :O. Maybe I will have to scooter over to hw and see what these VIN values are.


The core clocks also fluctated below 4.0, is that normal?

 

[RaDiKaL_]

It's common for the cpu vcore to drop, but not that much, it should go from 1.4v to around 1.375~1.350 at worse, and no, your core clocks should go between 4.0 and 4.2 (if turbo boost is enabled), it should only go lower than 4.0 if you're at idle, not during gaming.

Clearly your CPU is throttling down itself, and only two things can cause that, overheating or lack of power...

One long shot is if your mobo's VRMs are overheating, that could also lead to the CPU throttling down but from what I see your mobo has a nice big heatsink on them, and as long as your CPU cooler blows air towards them then that shouldn't be a problem.

So yes... now there's more evidence pointing to the PSU, your new Supernova should fix these issues.

Now it becomes more

 

[KinesticBat]

Will update this thread when I receive and test it

----------
Update:
----------

After replacing the PSU, I'm happy to say I'm surprised and optimistic. I noticed many differences. Today I was able to do some good comparative readings. On idle, the CX vCore would fluctuate between ,8V ~ ,1.2V and on load it would present vCore dips sporadically during gameplay. Power (CPU wattage) would spike and often dip low during stress testing (dips from 55W to 30W for example).

The G2 on the other hand keeps a strong power flower, staying only within the range of ~50W, never deviating more than +/- 10%. The vCore is spot on. On idle it is a constant ~1,3V - ~1.4, the same with gaming, and any load. Input CPU voltage has increased just a little from 1,.808 - 1.848 (maybe it was hungry?). The mysterious VINs have not reported any drops. So far it is looking to have been a power issue. The maximum my system draws is ~530W. I suspect the CX was not putting out even that much or was unstable on a rail. Chassis temperature has also improved. Will give this a good week for more result :d

The result will be definitely determined with ME:A around the corner, but I have high hopes of these results. Oh and yes, no freezing thus far xD

-----------
Update 2:
-----------

I consider this solved and to have been a case of a failing / unstable power supply as a result of component upgrades. 😀
I have had no crashes for the past few days and power consumption has gone up, point that the GPU and CPU were probably being starved D:
Thanks, RaDiKaL!