BSOD when I restart inside of windows

goldfish · 29 May 2017

i7-7700k
Asrock z270 Extreme4
32GB DDR4-2400 Ram
GTX 1080Ti FE
Corsair RM750x
Samsung 960EVO 500GB M2
Seagate barracuda 6tb x2
Windows 10 64bit Build 15063

I've recently built, delided and overclocked my first pc. Everything was running just fine, until recently, I found out that pressing restart inside the windows start menu will shutdown windows, but will always freeze when it starts up again on the spinning dots. It freezes even if I overclock by 100Mhz. It freezes when I manually set multiplier to 45 even though its a 4.5Ghz cpu. The pc starts up just fine, runs just fine, and shuts down just fine. This only happens when I use the restart option inside windows or when an update causes it to restart itself.

I have my cpu OCed to 4.9Ghz at 140mV offset LLC3. Under load it's doing 1.28v. Temps never hit over 79C in p95. My OC is stable for 10 Hours p95 blend, 11 hours p95 small FFT, and an overnight memtest86. It has never crashed on me inside of windows.

Googling tells me that the cpu can't handle the overclock, but I'm passing these stress tests with flying colors. Not only that, but manually setting the cpu to 4.5Ghz results in the same lockup when restarting inside windows. This doesn't happen when I set multiplier to 44 or below, or on stock settings. I've tried updating BIOS from 1.4 to 2.0. I've tried reseating the cpu. Nothing works. Most times, it would restart, and get stuck on the spinning dots and do nothing. Sometimes, it would BSOD after getting stuck on the spinning dots (either WHEA uncorrectable or machine check exception) but the pc will hang at 0% as it's collecting an error report. As such, I'm not sure if my dump log is of any use, but I've attached it anyways. I would appreciate if anyone could help me out or share your past experiences if you have resolved this before on your pc.

Attachment 137239

BSODHunter · 29 May 2017

That would be something for an overclocking forum. I think u will get better help there. But I was interested and analysed for u.

These are the results:

Code:

0x0000000A:    IRQL_NOT_LESS_OR_EQUAL (16.05.2017 11:51:05) [Windows 10] 
 
CAUSED BY:     ntoskrnl    
 
PROCESS:       System 
 
Usual causes:  Kernel mode driver, System Service, BIOS, Windows, Virus scanner, Backup tool, compatibility

0x0000000A on an OCed System:

Code:

0x0A = unstable RAM/IMC, increase QPI first, if that doesn't work increase vcore
0x0A = add QPI/VTT voltage

read this: Common BSOD Error Code List for Overclocking - Overclock.net Community

The GPU driver also crashed very often:

Code:

2017-05-26T01 36:16.996  Display driver nvlddmkm stopped responding and has successfully recovered.: . 
2017-05-26T01 36:16.997  Display driver nvlddmkm stopped responding and has successfully recovered.: . 
2017-05-26T01 49:09.063  Display driver nvlddmkm stopped responding and has successfully recovered.: . 
2017-05-26T01 49:09.063  Display driver nvlddmkm stopped responding and has successfully recovered.: . 
2017-05-26T02 04:10.159  Display driver nvlddmkm stopped responding and has successfully recovered.: . 
2017-05-26T02 04:10.159  Display driver nvlddmkm stopped responding and has successfully recovered.: .

But i cannot say if the driver crashes because of the overclocking or has another cause.

Last word from me: MSI Afterburner is known for causing bluescreens too.

goldfish · 29 May 2017

Thanks for taking the time to help me! I will make a post over there as well. I tried googling ntoskrnl and the issue could be memory. But I've already ran memtest86 overnight with 0 failures. I've also tried bumping up the voltages for SA and VCCIO.

I've looked at the file dates for those dumps, and the date modified is showing 5/16. I'm not sure how much use those dump logs are. However, I've taken a look at some of the files the dm log collector made, and some of the event logs are showing today's date. Not sure if that data was taken from the dumps, or from my current configuration.

goldfish · 29 May 2017

I found a fix for this! Originally, I ran into a fix where I changed VCCSA and VCCIO, while giving it a slight bump in vcore and changed cache speed (ring clock) to 4.8Ghz. This allowed me to boot and restart!!!!!! But it would instantly crash the moment I ran a stress test... So I had a computer that was super stable and couldn't restart, and now I have the exact opposite.

After a bit of testing, I found that changing VCCSA and VCCIO to anything other than stock would cause instability. The key to a stable boot and restart was cache speed. Setting the cache speed to anything higher than stock would require more voltage, so I wanted to see how low I could set this without causing instability. Turns out I only needed to set the cache speed multiplier to 42, the same speed as if it was left on auto... Completely bizarre.

tl;dr Computer is very stable but crashes on restart. Cache speed if left on auto clocks at 4.2Ghz. Setting cache speed to 4.2Ghz fixes restart problem.

BSODHunter · 29 May 2017

Glad to hear that your problem is now fixed. Let us know if there are any other problems.