Question Constant but inconsistant reboot during games, need help

[ejelliott17]

Running Ryzen 5 3600 with stock cooler, ASUS TUF Gaming x570+ (on current BIOS Version 2607), 16GB g.skill Ripjaws v series (3600MHz but bios set to 2133), PowerColor RedDevil RX 5700 XT (current drivers 20.8.2), and EVGA Supernova 650 G+ (80+ Gold), 3 Corsair AF120 in suction: DC mode Chasis fan1-3 (silent, standard,turbo - top to bottom), 2 corsair af140 on CPU_OPT (DC mode, 55%under 30C, 75%under 65C, 99%@ 75C), Samsung 970 Evo 1 TB M.2-2280 NVME Solid State Drive. I have windows 10 updated completely to today's date. I even did a clean re-install last week, reinstalled chipset drivers, gpu drivers, sound and LAN. I consistently crash in COD MW. No error codes just black screen reboot to login. I have run the game at minimal to high settings. GPU temps are under 76, utilization fluctuates between 30 and 99 depending on circumstances. Reboot almost every time I play. I have had reboots in Witcher 3 and Apex. I have dropped all settings to low and still had complete reboots. I have read at least 50 different, change this windows setting, adjust your power profile, type D3D11 in the extra command lines, run as administrator, change the memory scaling to .5... Nothing seems to work. I have never had a crash/reboot while NOT gaming so windows seems to be functioning properly.

If anyone has any advice I would love to hear it.

 

[Grobe]

Ok, do this steps first:

• Install OCCT and perform a stress test while it draws a time/voltage curve. Post a screenshot of that.
• Also have a run with Memtest to see if ram is good.

 

[michaelnaber]

me too

 

[ejelliott17]

Ok, do this steps first:

• Install OCCT and perform a stress test while it draws a time/voltage curve. Post a screenshot of that.
• Also have a run with Memtest to see if ram is good.

Memtest ran fine. No problems with RAM. I have also ran Cinebench and Unigine 4.0 on max settings. No crashes. I have OCCT up and ready to test but I have a questions about the plot settings. It auto selected some "nodes" (for lack of a better word) and didn't select MANY others. Should I run with their auto selections or do i need to make sure they are all selected. Screen shot is before any test has been ran.

 

[Grobe]

Referring to this thread:

Question - Adaptive/offset voltage mode

So basically if I set voltage mode to Adaptive, is the voltage going to go up and down based on the cpu load, but never go above the vCore - is that right? Also, my understanding of Offset mode is this: The vCore is going to be the maximum load voltage (as in the case of adaptive mode) and the...

forums.tomshardware.com

Do you have a setting in your bios that says adaptive voltage or something? Try to deactivate, and then run same test over again.

If that isn't the case, then you may have a voltage issue - but the VIN8 seem to be regulated somehow - so unless OCCT it's just OCCT having a terrible slow update rate, and if latter, it may still indicate a voltage issue still.

 

[ejelliott17]

link below is three quick tests (10 min or less). occt _Large data set w/o AVX, LINPACK, OCCT small data set w/AVX. Tried doing the GPU test, shader complex. 3, usage cap 100% and it crashed immidiately.

About the adaptive voltage, after looking at the more stable results in the link, do I still need to look for this setting?

OCCT CPU tests

 

[ejelliott17]

UPDATE: I tried re-running the GPU test at 80% and again, 2-4 seconds in black screen to system reboot. I tried the OCCT MEMTEST screen shot below. Made it to 12 iterations with something like 278000 errors detected. Ran the POWER TEST and got maybe 1 minute and again, black screen system reboot.

This is very discouraging! First build man maybe 3 months old!!

 

[Grobe]

Do you happens to know what fan the AUXFANIN4 graph represents ?

Can you have a loo into Windows event viewer (relevant article on howtogeek) and see if there is any records that matches the time of crashes ?

 

[ejelliott17]

I do not know which fan that is off hand. I will single it out and let you know.

The three 120mm fans are on mobo cha_fan1 (top front), cha_fan2 (middle front), cha_fan3 (bottom front), 2 140mm are on a splitter plugged into cpu_opt, cpu fan in cpu_fan, and i have the stock coolermaster fan in aio_pump. all are in dc modes (they are not PWM fans) . Front fan profiles top to bottom are set at silent, standard, turbo setting currently (air flow was bypassing GPU area so i wanted to speed air intake at the bottom and slow it down at the top. Both 140mm fans are on standard profile.

Windows event view has tons of errors. Some around the time of the previous crashes and some since I got back and turned everything back on. Hard to pin point. Since turning the machine back on and running the GPU/MEM/POWER test there are 2 winlogon_information events, 17 windows error reporting_information events, 17 ESENT_warning events, and 2 application error_ERROR evnets within the WINDOWS LOGS>application.

Also, I tried to recreate the reboot in the GPU,MEM, & POWER test and ran 2-4 minutes in each with no shut down and no errors. I will check the logs next crash I can produce.

 

[ejelliott17]

Since the last message I have ran dkschk, mem diagnostics and combed through the event logs. Everything seems to point at a faulty PSU. Guess I will go down that road with EVGA and see what they can offer.

 

[Grobe]

Everything seems to point at a faulty PSU.

Well, that makes sense.

However, be aware that the 1-1,5V is not provided by the PSU but rather by a voltage regulator on the mainboard.
https://www.smpspowersupply.com/connectors-pinouts.html

 

[ejelliott17]

been a while but i spent a lot of time in the event viewer. everything was pointing toward PSU. kernel event 41 or something. so i borrowed a friends 1000w psu and haven't had a crash since. i image 1000 is overkill but def changed the game for me. thanks for pointing me in the right direction.