Question PC crashes during games. Black screen and fans maxed out!

[Jblaze24]

Hi i am pc noob trying to troubleshoot gaming pc so any info will help.

PC crashes when gaming but works fine when browsing web, streaming movies and everything else.
When i first start playing each day it will play for maybe hour then crash. Black screen and fans maxed out. pc stays on and i can still hear the game like normal. I have to hold down power button to restart.

After the hour when i first get on and it crashes that 1st time it will consistently crash within 5 minutes after restarting pc and playing game.
Seems like over heating since it takes awhile to crash then keeps crashing but the temps are fine as far as i know. Unless the max temp cutoff is set really low on CPU or GPU? MSI afterburner says max temp is 84 for GPU. Had the rig 2 years no problems till this started slowly about month ago.

I check temps during game with hwmonitor and CPU is 50 degrees and GPU is 65 degrees
i uninstalled and reinstalled windows and nvidea and updated all drivers so don't think it software because its not crashing at start or randomly. Only after i have been playing for like hour then keeps going black screen and maxing fans.

I upgraded PSU from 650 to 750 and still crashing same exact way. so just trying to troubleshoot but like i said i am pc noob.
I guess i just have to replace 1 component at time but don't really know where to go because the temps are fine as far as i can tell on CPU and GPU.
I ran CINEBENCH 30 minute stability test on CPU 100% maxed to try to make it crash like it does during games but it completed test without crashing. Guessing since it didn't crash i can rule out CPU? The game only runs 40% CPU. I'll post the results below:

Mainboard Max temp: 32
CPU Max temp: 91
CPU MAX package temp: 94
CPU Max core temp: 94
CPU Max voltage: 1.241
CPU Core Clocks: 4800MHz
SSD Max temp: 33

Graphics card had 3 different temp listings: GPU: 51 Memory: 61 Hotspot:63 (sidenote: GPU hotspot does get to 80 degrees when it starts crashing but the GPU temp stays in between 60-65 on both Hwmonitor and afterburner.

System: 8core i7 9700k 3.6GHz No Overclock RTX 2080 super
CINEBENCH score: 8626

I don't know anything about numbers or temps just posted the results to try to help but any info would be appreciated.
Like i said i'm pc noob but common sense tells me since it made it through the 30 minute test maxed out at 100% that it is probably not the CPU. It does have new thermal paste too. Just don't know where to look next. PC also crashes on other games.
Is there way to stress test graphics card?

 

[Phaaze88]

Why would the voltage need to change? Don't you think they set voltage strong enough for some wiggle room for people who they know will OC? If That was the case then no one would be able to OC.

Unlike the gpu core voltage, which changes dynamically and you have some degree of control through software(Ex: Afterburner Curve Editior), the voltage going to the memory chips is static and control is locked.
If it weren't locked, users could push higher Vram OCs.


OC is pretty dead - no, I take that back. The gate of entry is just really high for most people; if not using LN2 or custom liquid + bios modding/shunt modding, one is quickly stopped by thermals, integrated power and voltage limits, or a combination of them.
Intel cpu: can't push them nearly as hard compared to like 10 years ago, because Intel already does the OCing for you. Some users are continuing to do it with the newer cpus anyway, doing what I like to call, 'sacrificial OCing'; giving up single core performance, which is important in so many applications out there, for all core performance, which has more limited use.
The cores' max turbo also scale up/down based on how many are active at a time... there's a lot of scenarios where folks have broken even or lost 100-200mhz on all core OC, and not be aware of it...

AMD cpu: These guys function quite a bit like Nvidia's gpus do - at least Ryzen 3000 and 5000 do - dynamically adjusting boost based on temperature and power use. Most of the work is done for you.
Can get a sliver more through the use of Clock Tuner and Dram Calculator, if one is patient.

Nvidia gpu: Gpu Boost algorithm does the majority of the OCing(gpu core only) for you, and dynamically too. If an aftermarket model was purchased, that's maybe a couple more % on top of that.
The board power limits are very conservative and voltage control is limited on models before the 10 series. I guess they were seeing too many reports of damaged cards from OCing.

AMD gpu: Haven't really looked into them, but I believe they have a function similar to Gpu Boost. Again, most of the work is already done for you.

Spoiler

View: https://www.youtube.com/watch?v=wnRyyCsuHFQ

It's a 20 series, but I'd say some of it still applies.
You already tried changing the psu to a good one, with no luck.
Used DDU.
Reinstalled Windows.
The gpu may be it after all...

 

[Jblaze24]

Phaaze88 said:

It's a 20 series, but I'd say some of it still applies.
You already tried changing the psu to a good one, with no luck.
Used DDU.
Reinstalled Windows.
The gpu may be it after all...

It is the VRAM memory chip speed going to slow causing the crashes. or something inside the GPU wearing down causing it to go slow like you mentioned.

Either way it's good for now and I know what the problem is at least.

Just thinking though, I think your theory about components wearing down might be right because there are a few videos I remember seeing on YouTube that was having exact same problem and fixed it. It was connection inside GPU. They could replicate the problem by shaking PCI-e cable.

I ruled that out because I tried shaking the cables and couldn't get it to happen. It does make sense because I had a few interesting results testing MSI Afterburner doing 3Dmark . I will post my 3D Mark results from Afterburner along with the link to the YouTube video. The problem in the video is the exact same problem I was having, also with the exact same error message which I had no clue what it meant even after searching online. The error message is Kernel power 41

3D Mark test results with MSI Afterburner

Test 1: 2 Initial runs with no changes. Crashed at 15 seconds both times
Test 2: Decrease core clock -200MHz. Crashed at 15 seconds
Test 3: Decreased memory clock-200MHz. Crashed at 15 seconds
Test 4: Increased power limit to 120%. Crashed at 8 seconds
Test 5: Increased core clock +200MHz crashed at 31 seconds
Test 6: Increased core voltage from 0% to 100%. Crashed at 40 seconds
Test 7: Increased Memory clock MHz+500. No crash
Test 8: Tested 3D Mark 3 more times and played 10 games with no crash. Problem fixed

Here is the link to the YouTube Video I mentioned:

ASUS RTX 2080 Power Connector Repair (Black Screen Crash/Loud Fan Fix) (Mini Projects Ep.9)
30,561 views
Nov 14, 2020


Fuzzy Dice Projects

265K subscribers