Ryzen Overclock Causing Windows to Crash?

[Reven]

Curious OC problem. I'm running a 1700 with default cooler and the ASRock Taichi mobo, Samsung 960 evo and GTX 1080 TI.

Currently I have the 1700 OC'd to 3.8ghz at 1.37 voltage. Everything seems fine (played games for 2 hours, no instability).

I've been following this guide to OCing: https://www.youtube.com/watch?v=52Tw-wcT7o4

What I find strange is when I start playing with the custom settings, Windows refuses to boot. I get a an error saying "Windows failed to load because the kernel is missing, or corrupt: 0xC00000e9 – File Windows\system32\ntoskrnl.exe."

At first I thought it was a genuine OS issue and did a repair, but it keeps coming up anytime I play at higher clockspeeds. If I change the bios settings it is totally fine.

Why would CPU overclocking trick Windows into a kernel failure?

 

[richierich1212]

What I find strange is when I start playing with the custom settings, Windows refuses to boot. I get a an error saying "Windows failed to load because the kernel is missing, or corrupt: 0xC00000e9 – File Windows\system32\ntoskrnl.exe."

What custom settings are you talking about?

 

[Elixer]

Why would CPU overclocking trick Windows into a kernel failure?

This isn't Ryzen specific. It happens on intel, nvidia, via, and pretty much anything that can be O/C'ed.

This can happen on ANY unstable o/c. Which is why overclocking is never guaranteed to work.

Could be the CPU, RAM, voltages...and so on.

 

[Reven]

@richierich - it was editing the pstates.

From the youtube channel:

"Values for P State Overclocking -
FID (Clock Speed) 90 - 3600mhz, 91 = 3625, 92 = 3650, 93 = 3675, 94 = 3700, 95 = 3725, 96 = 3750, 97 = 3775, 98 = 3800, 99 = 3825,
9a = 3850, 9b = 3875, 9c = 3900, 9d = 3925, 9e = 3950, 9f = 3975
a0 = 4000, a1 = 4025 (my max sweetspot OC), a2 = 4050, a3 = 4075, a4 = 4100, a5 = 4125, a6 = 4150, a7 = 4175, a8 = 4200 = a9 = 4225, aa = 4250, ab = 4275, ac = 4300, ad "

 

[John Carmack]

Everything seems fine (played games for 2 hours, no instability).

Why would CPU overclocking trick Windows into a kernel failure?

You need to test stability with more than just games.

 

[DrMrLordX]

You would be surprised by the number of different failure types you can have from overclocking. Sounds like you are somehow pushing the IMC out of spec. Try raising SoC voltage and see if that doesn't help some. And maybe VDDP.

 

[Despoiler]

I recommend using x264 stability tester. It's a real world workload and it will pull out any OC instability. Gaming testing is not a stability test.

 

[DrMrLordX]

Game testing only tests stability in that particular game. So it's valid for that purpose alone.

 

[Despoiler]

Game testing only tests stability in that particular game. So it's valid for that purpose alone.

Sure, but using tests like x264, prime95, LinX, OCCT, etc will allow you to get stability for everything. For example I can run my chip @ 4GHz @ 1.33v for games that have low CPU utilization. Immediate black screen for anything like BF1, rendering, or crunching workloads. My actual stable for 4GHz is at least 1.41v. I haven't bothered to find it because it's more voltage than I'm comfortable with. I settled on 3.95 GHz @ 1.37v. To me it seems pointless to test with a game or games when you can use a single program to find true stability. Each to their own though.

 

[Rifter]

Sure, but using tests like x264, prime95, LinX, OCCT, etc will allow you to get stability for everything. For example I can run my chip @ 4GHz @ 1.33v for games that have low CPU utilization. Immediate black screen for anything like BF1, rendering, or crunching workloads. My actual stable for 4GHz is at least 1.41v. I haven't bothered to find it because it's more voltage than I'm comfortable with. I settled on 3.95 GHz @ 1.37v. To me it seems pointless to test with a game or games when you can use a single program to find true stability. Each to their own though.

Completely agree, its either completely stable, or its not. And i dont settle for not. I settled at 3.9 for the same reason as you, i wanted 1.35v as max voltage, and that got me to 3.9

 

[DrMrLordX]

I don't disagree, it's just that for some gamers, they'll have a tune specific to games (or even a particular title). They don't really care if their rig is AVX stable 24/7 or whatever. They just want it to run Dota2 at 144 fps or 240 fps or . . . whatever.