PBO causes WHEA errors on Ryzen 5 5600

[Juhrien]

I've been struggling with an issue for the past few days.

When PBO is enabled and I launch any demanding game, I get a WHEA_UNCORRECTABLE_ERROR blue screen after 1-2 minutes of gameplay. Before I realized the issue might be related to PBO being enabled, I went through a long list of tests. I updated all my drivers(even LAN drivers), checked the disks with HD Tune, analyzed the dump file, ran memtest86, uninstalled recently installed apps, basically tried almost everything I could think of. I figured out the issue was caused by PBO through just messing around with BIOS settings and running some tests.

The strange part is that the system runs completely fine under stress tests like OCCT or Prime95 Small FFT. The problem only happens during actual gameplay. Yes, disabling PBO does solve the issue, but what confuses me is that even with PBO enabled, the system doesn't crash during OCCT or Prime95 Small FFT, it runs without any errors for 10–15 minutes(didn't feel the need to test it any further) straight.

I should mention that during Prime95 Small FFT with PBO enabled, my CPU hits around 90°C, whereas it stays around 60°C during OCCT. Also, when PBO is disabled, even Prime95 Small FFT runs around 60-65°C. I do have a proper cooler, and I don't think this is a thermal issue, because when the system crashes during games, my CPU temps are only around 60-65°C.

I’ve attached the dump analysis to the attachments.

 

[DirtyDingusMcgee]

System specs please.

You've answered your own question. You'll need to tweak voltages or adjust pbo curve and power limits. could be quite a few things. With no system specs or idea what you've messed with in the bios, it will be hard for anyone to help.

 

[Juhrien]

System specs please.

NVIDIA RTX 4060 Ti 8 GB
AMD Ryzen 5 5600
GIGABYTE B450M S2H
Kingston Fury DDR4 32 GB (2x16) 3200 MHz
Kioxia Exceria Plus G2 1 TB M.2
Aerocool VX Plus AE-VXP500 500 W Power Supply

You've answered your own question. You'll need to tweak voltages or adjust pbo curve and power limits. could be quite a few things.

I want to figure out why I'm running into this issue when PBO is set to auto.

With no system specs or idea what you've messed with in the bios, it will be hard for anyone to help.

Like I said, the issue goes away when I disable PBO, and comes back when I turn it back on. So it's clearly tied to PBO being set to auto. But why? That's what I want to figure out.

 

[DirtyDingusMcgee]

No mention of your cooling in specs. Power supply is garbage. That could easily be the issue.

PBO set to auto or set to enabled? Not the same thing.

Make sure you are running latest bios (there are like 23 bios versions for that board). Then load defaults for motherboard. Don't change anything except turning on pbo....see if the error persists. If it does, you'll need to tweak power. If issue doesn't arise, then tune your memory (xmp i am assuming) and try testing again.

 

[Juhrien]

No mention of your cooling in specs.

It's not a brand you'd know since it's only sold in my country, so I didn't mention it, but just so you know, it exists.

Power supply is garbage. That could easily be the issue.

Then why doesn't it cause issues under load? My CPU runs with 140W without any problems.

PBO set to auto or set to enabled? Not the same thing.

Auto.

Make sure you are running latest bios (there are like 23 bios versions for that board). Then load defaults for motherboard. Don't change anything except turning on pbo....see if the error persists. If it does, you'll need to tweak power. If issue doesn't arise, then tune your memory (xmp i am assuming) and try testing again.

My BIOS is up to date. I've already tried resetting to defaults, but it didn't change anything. PBO is enabled by default. Like I said, even when PBO is on, this issue only happens in games.

 

[A Computer Guy]

I want to figure out why I'm running into this issue when PBO is set to auto.

When PBO is set to auto it hides other options like PPT, Scaler, Curve Optimizer, etc... correct? In the UEFI/BIOS I would try setting PBO to advanced, PBO Voltages (for PPT, TDC, EDC) to AUTO, Scaler to manual 1x, +200MHz, CO disabled. Avoid using Ryzen Master to change settings.

Then why doesn't it cause issues under load? My CPU runs with 140W without any problems.


My BIOS is up to date. I've already tried resetting to defaults, but it didn't change anything. PBO is enabled by default. Like I said, even when PBO is on, this issue only happens in games.

Games are boosting most of the time not all core workload so when boosting its trying to hit the higher frequencies it pulls the most voltage per core for a very short amount of time like pressing on the gas pedal to accelerate a car. UEFI could also be buggy.

Like I said, the issue goes away when I disable PBO, and comes back when I turn it back on. So it's clearly tied to PBO being set to auto. But why? That's what I want to figure out.

Do you have LLC options on your motherboard BIOS/UEFI? If so then try adjusting LLC from the lowest setting to a midrange setting to counteract possible voltage drop. Maybe when PBO is set to auto your not getting the right voltages when the CPU boosts.

When PBO is enabled and I launch any demanding game, I get a WHEA_UNCORRECTABLE_ERROR

I don't recall if that's a memory error. You can try increasing your SOC voltage to 1.1v (or 1.2v maximum)

 

[Juhrien]

When PBO is set to auto it hides other options like PPT, Scaler, Curve Optimizer, etc... correct?

Yeah, doesn’t show up unless I set it to Advanced.

In the UEFI/BIOS I would try setting PBO to advanced, PBO Voltages (for PPT, TDC, EDC) to AUTO, Scaler to manual 1x, +200MHz, CO disabled. Avoid using Ryzen Master to change settings.

I'll give it a try and share the results.

Games are boosting most of the time not all core workload so when boosting its trying to hit the higher frequencies it pulls the most voltage per core for a very short amount of time like pressing on the gas pedal to accelerate a car. UEFI could also be buggy.

Since the tests are synthetic, they can't really simulate things the same way games do so you're saying it might be because games cause sudden voltage spikes or similar behaviors?

I don't recall if that's a memory error. You can try increasing your SOC voltage to 1.1v (or 1.2v maximum)

From what I've researched, this error code usually points to a hardware component causing a failure. I recently ran a MemTest86 check my RAMs.

 

[DAPUNISHER]

Power supply is garbage. That could easily be the issue.

That's the prime suspect alright. Next on my list would be too aggressive out of box settings causing the CPU to degrade. Every AM4 4 and 5 series board I've used had aggressive out of box settings. Be it ASRock, MSI, Asus, Gigabyte. CPU voltage would max at 1.5v+ which is excessive IMO. Especially when the CPU will run the same speeds or better with ECO or a negative CO.

 

[neatfeatguy]

Then you also have the plausible reason that you may not be able to run PBO without this problem of crashing.

I've had GPUs that would overclock well, be stable in any stress test, but then when I played a game they would crash. Tweaking voltages and making sure temps were optimal didn't help, they just weren't stable in games. Same has gone for CPU overclocks, things run stable in stress tests, but when doing other intense tasks such as gaming or transcoding it would become unstable.

Just something to keep in mind. Best of luck on your journey.

 

[Juhrien]

When PBO is set to auto it hides other options like PPT, Scaler, Curve Optimizer, etc... correct? In the UEFI/BIOS I would try setting PBO to advanced, PBO Voltages (for PPT, TDC, EDC) to AUTO, Scaler to manual 1x, +200MHz, CO disabled. Avoid using Ryzen Master to change settings.

Can't see anything about these settings. Can you walk me through it step by step?

 

[Kodehawa]

Honestly FYI PBO is pretty much overclocking (but modern). You'll need to tweak stuff like CO, it maybe needs a bit more voltage if you want to get it stable.

If system is unstable with PBO I'd just leave it stock, probably you just got a lemon but at least it runs fine like that.

Can't see anything about these settings. Can you walk me through it step by step?

View attachment 401128

Set PBO limits to 120/60/90 (PPT, TDC, EDC), you'll see it once you click "Advanced" or "Manual on PBO Limits. Set Precision Boost Overdrive Scalar to 1x, do not touch CPU clock override (Set it to 0! it's now 25, and the instability might just be that), do not touch CO (set all to 0 if you have touched it before).

 

[freeagent]

-25 on stock power is your problem.

I run my 5600X at 200/140/180 +200 -25 and have it boosting to 4800.

 

[Juhrien]

-25 on stock power is your problem.

I run my 5600X at 200/140/180 +200 -25 and have it boosting to 4800.

-25 is the default when I set that PBO to Advanced. It's probably getting changed automatically when it's set to Auto.

Honestly FYI PBO is pretty much overclocking (but modern). You'll need to tweak stuff like CO, it maybe needs a bit more voltage if you want to get it stable.

If system is unstable with PBO I'd just leave it stock, probably you just got a lemon but at least it runs fine like that.

Until last week it was working fine in PBO auto. I have been using this CPU for more than 2 years. It just started happening all by itself.

Set PBO limits to 120/60/90 (PPT, TDC, EDC), you'll see it once you click "Advanced" or "Manual on PBO Limits. Set Precision Boost Overdrive Scalar to 1x, do not touch CPU clock override (Set it to 0! it's now 25, and the instability might just be that), do not touch CO (set all to 0 if you have touched it before).

What's next? Should I do stress tests before jumping into games?


Edit: I tested it in game, but nothing changed. I still got a crash after 1-2 minutes. I had Core Temp open on my second monitor and kept an eye on it. I noticed that the VID value in Core Temp kept fluctuating throughout the game, between 1.16 and 1.18 (usually staying around 1.18). Is that normal?

If it helps, I can screen record some values for about a minute while playing the game, in case anyone like to see any specific values.

 

[freeagent]

Take note of your 12v, 5v, and 3.3v rails when you get the crash.

 

[Juhrien]

Take note of your 12v, 5v, and 3.3v rails when you get the crash.

The 20 seconds before the crash, the screen goes black right before it happens. https://streamable.com/30hjqv
Edit: I changed the update interval to 100ms and recorded a new video https://streamable.com/4fh2dm . It shows the last 30-35 seconds before the crash. At the end of the video, the system crashes and the sensors freeze.
Note: About 1.5-2 seconds before the crash, the screen freezes and the audio stops, then the blue screen appears.

 

[freeagent]

Yeah the 3.3v is the cause of your crashes. Maybe try reseating your connectors, 12v and 5v look pretty solid so it could be a fitment issue somewhere.

 

[Juhrien]

I bought an NZXT C750. I'll share an update on how it goes in the coming days.

 

[Juhrien]

I bought an NZXT C750. I'll share an update on how it goes in the coming days.

I installed it in my PC today. After a clean installation, I'll test it by running some games to see how it performs. But for now, the voltage seems stable under load.

(A screenshot of the values under load, not much fluctuation)

By the way, I'll go with Windows 11 IoT Enterprise LTSC 24H2. I'll be doing an unattended install and setting up Atlas OS. What do you think?

 

[RJARRRPCGP]

Could be a poop motherboard. I usually myself would just go to a decent B550 motherboard.

Also, if the graphics, Unigine Superposition 1080p Extreme should fail like clockwork, if unstable. Often, entire PC crashes.

Often, Maxon Cinebench will reveal core instability. Usually, PC crashes major time, if unstable. (Or just suddenly turns off, but that's 99 percent because of a critical overheat)

 

[anonuser57]

It's possible that you're mobo doesn't keep up with rapidly changing power demands too well. Changing your LLC settings can help with that. I have also had a motherboard behave that way because VRMs were dying. That you were putting 140W into your CPU at some point is concerning with that sort of board.

 

[RJARRRPCGP]

Heck, even my 5600X, I doubt would need to draw 140W! This ain't a "regular" 5800X we're talking about or a 5900X!

(Especially the non-3D 5800X, would draw 142W like no tomorrow.) (Not a big deal on 5900X with a decent B550 motherboard, because they run cool, even at 140-something watts.)

Looks like that motherboard needs to get retired.

 

[blinnbanir]

I guess that is a reason that the 5600 is not called a 5600X.

 

[freeagent]

Heck, even my 5600X, I doubt would need to draw 140W! This ain't a "regular" 5800X we're talking about or a 5900X!

At 4700MHz with Linpack my 5600X can do 155w PPT lol.. my 58X3D is locked to 143w PPT, my 5900X can do 260w PPT, and so can my 9900X.

Linpack loads though.. nothing else I have ran gets em that high. Everything but the 5600X was with extended power limits, +200MHz, and my curve.

 

[RJARRRPCGP]

At 4700MHz with Linpack my 5600X can do 155w PPT lol.. my 58X3D is locked to 143w PPT, my 5900X can do 260w PPT, and so can my 9900X.

Linpack loads though.. nothing else I have ran gets em that high. Everything but the 5600X was with extended power limits, +200MHz, and my curve.

I never tried to take them to watt limits like that. I don't know if I can get better results at 155W PPT. I already was getting a lot of access violation errors with my 5600X, seems to possibly be a dud core OC'er!
Even when CTR back in the early-2020s, told me that my 5600X is a golden sample, LOL. OTOH, I don't even see 100W with my 5800X3D, usually, LOL. Is that because of the motherboard?
Does the Asus ROG Strix B550-F Gaming always run very-low watts for 3D by default? Will it be 142W if I pop it into an ASRock B550 motherboard?
I may have had it at 4800 MHz, when the access violation error occurred. (in a browser tab) I think I will be lucky to get 4700-4750 MHz for mine!

IIRC, +200 MHz would cause access violation errors with my 5600X.
I have to admit, I was scared to make major changes in the BIOS! I used CTR in the early-2020s, briefly.
I did manage to get a fatal WHEA for "Cache Hierarchy Error" when running Maxon Cinebench multicore at one point, IIRC.
Windows 10 rebooted itself and surely enough, it was a stop error. (Checked the event log)

 

[Juhrien]

That you were putting 140W into your CPU at some point is concerning with that sort of board.

Heck, even my 5600X, I doubt would need to draw 140W! This ain't a "regular" 5800X we're talking about or a 5900X!

(Especially the non-3D 5800X, would draw 142W like no tomorrow.) (Not a big deal on 5900X with a decent B550 motherboard, because they run cool, even at 140-something watts.)

Looks like that motherboard needs to get retired.

No, my CPU was pulling 140W during the Prime95 test. That test already ran solidly for about 10 minutes without no crashes. In games, CPU was rarely going over the TDP value anyway. I only included that info to point out that the CPU managed to hold up for 10 minutes even while using 140W. I didn't think it was related to the crashes. Anyway, like I mentioned above, the issue will probably be fixed with the new PSU. I still haven't had a chance to try it