Showing results for 
Search instead for 
Did you mean: 


Journeyman III

WHEA logger 18 - PC randomly crashing in games - Ryzen 5 3600X

I have this, as I saw, common problem with WHEA logger 18 crash on my PC. Here is the full spec:

AMD Ryzen 5 3600X 6-Core Processor (12 CPUs), ~3.8GHz
AMD Radeon RX 5700 8GB
Gigabyte Technology Co., Ltd. X570 UD
be quiet! Pure Power 11 600W
2x GOODRAM 16GB (1x16GB) 3600MHz CL17 IRDM PR

Crashes happen only in games and they are pure random. Sometimes I can have "safe" month or two, sometimes I get couple of crashes in one day. It mostly depends on how advanced is specific game (for example, Dirt Rally 2.0 is crashing often than Dying Light). I tried almost everything, almost every method with no effect. Let me tell you ALL I tried and all I know.


1. My PC, except this one issue, is working fine, with good temperatures, no overheating. It can handle stress-tests from various programs easily, nothing bad happening when doing that. I tried checking health of all components that can be tested and I never get even a single problem or damage.

2. I disabled "fast startup", reinstalled GPU drivers, updated every single driver. No effect.

3. I changed the following setups in BIOS

- PSS support/AMD Cool n Quiet -> disabled
- global c-state control -> disabled
- power supply iddle control -> typical current idle
- power down (DRAM) -> disabled
- gear down mode (DRAM) -> disabled
- core performance boost -> off

No effect.

4. I found the last solution I never tried before so I did that. Look - I set VCore to Normal and set voltage in the Differential field to +.006 volts. Same with VSoc and Differential field changed to +.006 volts. I heard that this crash can happen because of CPU idle voltage. And well... another no effect.


Please guys, help me I'm going insane!


44 Replies

Do you keep your Infinity Fabric at 1:1 to RAM clock (1800MHz)? I find it plausible that this might be the issue. Set FCLK to "Auto" and lower your RAM speed to 3200 (1600fclk) MHz. You may also try screenshotting / making photos of your 3600 RAM timings to then apply those for 3200, otherwise BIOS auto-timings might get too loose. 


Can you tell me how I can do that? I don't understand some definitions and also english isn't my native languague, so I am not completely sure about your message


Download Zentimings. 
Send screenshot from result of this application. 




They by results we may lead you to conclusion to either fix voltages to stabilize FCLK, or to reduce memory frequency a bit. 




Soo... you don't even have XMP enabled, so your RAM runs 2133 instead of 3600, and not even close to 3800. Fclk is obviosly normal (even though VDDGIOD is kind of too high for this frequency, can be 0.95V (950mV). Or IF you will use XMP it can be 1.0-1.05V (1000-1050 mV). 

VSOC is also adequate. 
You do use 2x 2R sticks, but you have 2T CMD rate enabled for that. Looks fine.
Had you tried with different RAM? 
Or had you tried to stress test your RAM? It may be faulty. You can use y-cruncher to hammer it with stress.
Try to download it and use stress test with these settings. 




Thank you for your respond. I don't have XMP enabled because when I did that I saw that I have a higher temperatures without any visible signs of better performance so I decided to back to default.

I am sure that my RAM is not causing this problem because previously I had 2 8GB RAM sticks and my PC was still crashing.

What can I do now? :(( There must be a solution


"I am sure that my RAM is not causing this problem because previously I had 2 8GB RAM sticks and my PC was still crashing."

With RAM frequencies over 3600 or while having more than 2 ranks of memory total (either 4 sticks ot 2 dual ranked ones) Ryzen memory controller may not be stable on 1T command rate. 2T can be more stable, and GDM is another stability step up. With cost of RAM efficiency, of course. So i may miss a bullet there, but WHEA 18 is memory controller error, and caused by SOC instability either in CPU or GPU (and there is a way you can edit SOC frequency and voltage on RX 5700 as well). On other side, from what you had written, there is no way you even know what MPT is, so GPU is likely out of equation. 

Strange that you had set 2T CMD rate by default on 2133 though. Can you link me original memory setup, as these crashes are irregular ones. So we could have seen values that are being set authomatically by your mobo for FCLK 1800. 



I'm sorry, but technical language + foreign language can lead to misunderstandings or mistakes, so I am not sure that I understood you correctly.

I checked in BIOS to see what I had in DRAM Controller Configuration in AMD CBS and AMD Overclocking - 1t or 2t. I had it on "Auto" so I changed to 1t because of two RAM sticks.

About original memory setup - what exactly do you mean? You want me to send you a picture of BIOS settings? If I understood you correctly, just tell me from which tab.


Nah, just boot in system with original setup (XMP and auto) and send Zentimings output as you did before with 2133 default. 

Maybe you had 3600 1T with quad rank setup.

"I checked in BIOS to see what I had in DRAM Controller Configuration in AMD CBS and AMD Overclocking - 1t or 2t. I had it on "Auto" so I changed to 1t because of two RAM sticks."

Number of sticks you use may not matter. You can use 2x 1R sticks (usually 2x4 or 2x8), 2x2R (can be either 2x8 or 2x16 as you have, or, maybe even 2x32 ), 4x1R (4x8 sticks) and 4x2R (4x16 or 4x32 sticks).
In order of increasing potential stability issues
2x1R ---> 2x2R ---> 4x1R ---> 4x2R.
But between 2x2R and 4x1R there are different issues. 4x1R usually have harder time clocking higher or getting tighter timings, because almost every mobo on market use daisy-chain RAM setup. While 2x2R have higher chance to get issues from getgo, but easier to work with in terms of overclocking. 

Ryzen Zen 2 and Zen 3 memory controller have tougher time dealing with anything except 2x1R from getgo if you are not lucky, so with 2x2R or higher there may be hard times booting into system on RAM clocks higher than 3200 (and for 4x2R it may be even lower) without turning on 2T or GDM (in order of increasing stability 2T ---> GDM)

Also, don't mix up RAM being dual-channel and being dual-ranked. 
2 channels mean that CPU memory controller is basically split by half, and each half works independently to reduce data latency and starvation by being able to alternate channels. It related to different slots on motherboard. if it is 2 slots, then each one have own channel. If it have 4 slots, then it is 1-1 2-2 if looking right on it, and usually marked as A2 A1 B2 B1 in motherboard instruction

Memory rank is related to RAM sticks themselves. Simply speaking it means how many sides on RAM stick is populated, 1 or 2. Why it matters? Because memory controller can ask only 1 side of 1 stick at once, so they need to switch "attention". And with more things to pay attention for, less attention averaged you can pay to one rank. That adds on over on top of frequencies. So if you have more than 1 rank in channel and run it on higher frequency (aka more than Zen 3 specced 3200) it becomes much taxing on memory controller. 


Okay, there are another screenshots.

Btw, I can't set "1t" in AMD CBS because after any reboot it goes back to "Auto". I don't have same issue in AMD Overclocking. And, mehh, I tried to play Snowrunner a bit and after 40-60 minutes I get a crash



Well, AMD CBS and AMD Overclocking are duplicating each other, so... Doesn't matter much
Command rate is usually changed from RAM timings page, and not CPU overclocking one. But if your way works as well, go ahead. 
But i still see that it is 2133 mbit/s, as there is no XMP enabled. And it still crashes? If yes, then you probably should stress test your RAM. It may be faulty.
Memtest86 for raw and medium precision memory test. Bootable from USB drive
y-cruncher with settings i sent above for high precision memory (and overclock overall) test. 
1 run is 100% NOT enough, you must pass at very least 1 hour without any error. 

Crashes also can be caused by GPU oc instabilities and/or application interactions (i have 2 specific games that cause BSOD in 15 minutes if both launched simultaineously). But as i already said before, from what i already seen from you i doubt you changed any clock related settings. 


Yes, I have still 2133 without XMP. As I said before, I decided to left RAM to default after getting higher temperatures and no better performance. And yes, it crashed again

I also tested every single component, also with Memtest. I tried running this program on old and new RAM sticks and nothing bad has been shown.

Randomness of this crashes are one of worst things about it. For example, I played Snowrunner for 6 months without crashes and now boom, it started to happen again. Three times in one week now.

I never, except first time with XMP, changed any clock related settings because I don't know much about it so I don't want to damage my PC.


Do you have more ideas? :((


I don't know why but when I try to edit last message I get "An Unexpected Error has occurred." so I must add question here, in a reply.

I am curious that, if default RAM clock (probably) causes crashes, XMP maybe will fix it? What do you think? Can it be a possible solution?


To be honest, highly unlikely. 
JEDEC specs on RAM are ones which must be consistently stable, while XMP is technically "auto" overclock. If RAM is unstable in stock, then it is more likely to be unstable with XMP enabled


Hmm, that makes it more complicated.

Do you have more ideas how to help me with this problem? I am really glad that you are helping a lot, not only giving me advices, but also telling me how this things really works, so huge thank you for that :))

As you can see, this is much more problematic as you probably thought at the beginning. It makes me completely nuts and you and this forum are my last and only hope.

What do you also think about Thomasdk1405 (thank you again!) reply from "technical" side? I used this method and I get about 2 hours of stability in Snowrunner, but it's much too early to celebrate


Well, newer versions version of AGESA can improve (or make it worse as well, for some people) memory compatability and/or stability. It can also improve or decrease performance in games/workstation apps. So this may actually work. 

But about mobo limits for PBO it is kind of BS. CPU can regulate this limits as well as mobo does. And some mobos can set really whacky limits making temps go out of control for some. For example 5600X have stock PBO limit 76W. But mobo limit for it is 88W as if it is 3600X (well i still run it up to 120W though, because i can). So setting limits from cpu to mobo can only change CPU performance if there is difference between presets. 

Also @Thomasdk1405 Only difference between mobo limits and cpu limits are... maximum values... sometimes. All voltages are requested by CPU and current is just straight up taken until limit. Power regulation doesn't have brain by itself. 


One thing is theory another is fact. In my case it worked setting this to motherboard and not auto. When back to auto it crashed again. So whatever the cause and reason for this - it solved the issue no matter what.

All I said was give it a try - and agree it should not be like this - should be auto all the way. I have read about people that RMA'd the cpu and motherboard several times and still they got this WHEA 18 and BSOD. In my case it was a brand new 5800 x and I was getting grey hair.

Set to motherboard it all stopped - and there has not been a single crash since then !


Sadly, this method wasn't a solution for me. Crashed again. Now I turned PBO off (previously had on Auto) and I hope it will do a thing.

When trying to fix this problem I spent many hours in many different sites. I heard that it can be caused by CPU clock differences when going idle and by CPU being auto overclocked by mobo functions like core performance boost etc. So I thought that making it as much stable as I can with stable voltages and clocking can, maybe, be a solution. I forgot only about disabling PBO. What do you think about that @DimkaTsv?


Well, auto OC is CPU feature, not mobo one. But it can cause problems when idling if voltage is low. Never had it myself as my sample is pretty good in undervolting.  
And also to be fair, if your CPU is unable to work under basic auto OC (which is activated from factory), then it is warranty/RMA case. Because these up to 4.X gHz clock boosts are in specs. 

But need to be sure it is AutoOC that causes issue, 
Btw PBO limit and AutoOC are bit different things. AutoOC is whole feature, while PBO limits is where upper limits of consumption lies for it.


It's too late for RMA because I am struggling with this problem almost two years (sic!). My PC was also in professional service and the CPU was sent to warranty. I get it back because "it's fine". Main problem is this goddamn randomness. IT specialists tested and stressed my computer in every possible way, using many programs - OCCT, Memtest, 3Dmark, HD Tune etc. No errors, good temperatures, no crashes.

This is completely insane situation. Even when I decide to give them my PC back, they won't have time to senseless play games on it (maybe for 1 hour, maybe for 1 month...) only to get a crash and see WHEA logger 18 in event viewer.

If turning off PBO won't help, can you maybe help me to safely underclock my CPU or gently raise the voltage on it? I can't do it by myself (I don't have enough knowledge) but I read on the internet that it can be a possible solution. I can "sacrifice" performance only to get a full stability


May it be come program incompatability? 
Or just driver error that is specific to game? 


I don't think so. It crashes randomly in random games. I can have couple of crashes in one day or zero by month or two. No dependency is visible.

Maybe some logs, codes from event viewer can tell more? Should I post it here?



Disabling PBO also wasn't a solution. Please, help me :((


Ngl, it is hard to judge what causes these crashes. 
CPU, RAM, GPU, software issue, literally anything. 
Wish we got more info on events before and after these crashes. History of actions, logs, event monitor,  BSODs (if there are any)... 


I can give you all info, just tell me what exactly you need


Games you got crashes (if multiple, which games have crashes often, which have them rare and which have singular time crashes) and what potentially causative applications you have running in background (monitoring, hardware control, another game, antiviruses, etc.)
What are you doing right before crashes happens (like, Alt+Tab or changing resolution, or attempting to screen record, etc.)

Driver settings (mostly if ReLive functionality is enabled)

How crashes usually go? Game freezes and errors out. Game silently goes to desktop. You get black screen and driver error. You get black screen and stuck game. You get black screen with sound but system is unresponsive. You get black screen and system reboots. You get black screen and then system goes in visual artifacts? 

Anything that can be exctracted from logs, dumps, event monitor etc... Right after an error happens. 


Okay, I'll try to answer you as best as I can.
The most "crashable" game was Dirt Rally 2.0, so I am not playing it anymore because it happen too often to have fun. The others most "crashable" are Snowrunner and, previously, Forza Horizon 4, but for some reason Forza "repaired" itself and Snowrunner worked very good for almost half a year. Sometimes, just like in my first days with this game, I got 3 crashes in two hours... but I also had month or two without any problems. It's just like last week - 1 crash per day and 6 safe days until today. Other games, like Dying Light or Dirt Rally 1 crashed 1-3 times only. Couple of another games never crashed, like Chronicles of Myrtana, PUBG, F1 2018, CS:GO, Mudrunner, Dark Souls, MK:XL. I guess it depends on how advanced is game, how hardware requirements look like. Strange, because my computer handles any game I tried on ultra settings without fps drops, and it also not overheating. Benchmark and stress tests never make my computer crash.
I don't have much programs working in background. It's just Steam, Windows Defender, AMD Software Adrenalin Edition and Mozilla Firefox.
When my PC crash, I am not doing any special or strange things. Just playing. Sometimes (in last year) I really had crashes when ALT + TAB, but, luckily, it no more happens. Crashes are, mostly, completely unexpected. Special effects, lights, shadows, explosions happening in games etc. are irrelevant. It can crash even when everything is calm and easy to calculate by CPU or GPU. Only in Forza Horizon 4 I had strange issue, because sometimes I was getting black screen when moving between locations (GB and Fortune Island).
Crashes now always look the same. I am playing and suddenly I get a black screen and terryfing, buzzing sound. But it's not a bugged or looped audio. It always sounds the same, no matter if game was silent or noisy. I hate this because its's much louder than crashed game and from time to time I even jump on my chair - I get really scared of this, lol.
I have this state by 2-5 seconds and my PC reboots and starts as always. No errors or notifications appear.
This is how it look in event viewer - (I used polish verion of pastebin because for some reason original one wasn't saving my texts). Most important is that everything look same in every crash... except APIC ID. This numer, for some reason, always is different.
And the details. This is what I can't understand so I hope you will -

If you need more info or logs/dumps, just tell me


Hmmm... What type of buzzing sound? It is likely sound like same sound repeated with high frequency (sound like trrrrrrrrrrrrrrr.... ), but not high enough to raise pitch?
I can just say that WHEA isn't necessarily straight up CPU error. 
It can also be caused by memory violation of some sort, even though rarely

APIC ID means logical core number error was recorded by. 10 means 6-th core main thread. (1-st core is 0/1, second is 2/3, third is 4/5 etc.). If this number is ever changing it means that issue most likely isn't CPU related (or you would've had much harder time dealing with it as these issues would become REALLY regular if every core is flawed)
RAM usually cannot DIRECTLY cause WHEA errors. But it can, when XMP profile is on and Zen based CPU Infinity Fabric (FCLK) cannot run at RAM frequency in 1:1 mode without voltage bump. But these errors are really frequent when happen, and can be caught by HWINFO very bottom row with hunderds of errors in minutes of load. And it doesn't necessary mean that workload will crash. But better be careful when seeing them. 

Anyways, if you passed about 3-4 hours of Y-Cruncher memory biased tests - you are pretty much fine. Also XMP actually enchances performance quite a bit. And higher performance lead to higher temps (up to some degree) as more data can be processed in same amount of time. Temp increase isn't usually high though. Don't remember my 5600X IOD chip heating above 40-45 degrees even when my CPU runs 80-84 (under heaviest loads).

And third and final cause i ever had leading to WHEA with unstable APIC ID is unstable GPU overclock. Especially... if i remember correctly GPU SOC frequency. And it sounds similar to what you have. You can also try to just downclock your GFXCLK on GPU (basically core frequency you can change through Adrenalin driver), as too high frequency can lead to very rare crashes (can be caught relatively often in something like God of War, and really often in CoD Warzone, especially if simultaneously recording screen)... You can also check GPU VRAM in case of potential erroring by checking it on artifacts via MSI Kombustor (1 hour of Furmark-Donut 6500 MB + fullscreen + Artifact scanning checked on)


It isn't solution, but it is steps that may help resolve issues and find culprit.


Yeah, this buzzing sound is very similar to "trrrrr". Something like that - - but louder, with lower pitch, little slower and lasting only couple of seconds. It is not easy to describe. This video is as closest as possible.

So, if CPU isn't related to these errors, what I can do now? I am still running without XMP profiles and my GPU isn't overclocked in any way. How (and how much) should I downclock GFXCLK on Adrenalin? I don't want to mess it even more by setting it all by myself. I will also try another benchmarks (those you told me about) but before I do that, I want to try every possible solution. Maybe testing my components more won't be needed if we fix this problem. Well... I hope.


Yup, normal sound of casual memory corruption or overload (when CPU cannot process all data in time causing sound jitter, but it goes away much faster, so i doubt this one is suitable for your case). As i expected. 

Technically IOD can cause this crap, but really unlikely, as if is tempted to corrupt data, it will happen much more often than once a week or day.

To reduce GFXCLK Just try to set maximum frequency about 100 mhz less than it is currently set for you (stock value). 


It is way to reduce gfxclk specifically. For SOCfreq and FCLK you need program called morepowertool... and... well... For you it's probably better not going deep into this one. 


Thank you for your help and response again. I will try this method next, because meanwhile I found and tried another possible solution. I am curious what pro like you will you think about it :))

One user had similar problems and he solved it by changing "Vcore load line calibration" from Auto to Medium, by setting Vcore from Auto to Normal (I did it before, also with putting volts to +.006) and by undervolting to -0,8125V. I am not sure what he undervolted (in Vcore I don't have similar values) so now I ignored this one. But the Vcore load line calibration looks, in definition, as very related thing to my problems with unstable PC, so I hope it will really help me.

I want also notice, that I have no idea which component is causing this problems. On other forums I heard that maybe RAM or hard drive, but when I bought new RAM sticks and after I changed HDD to SSD, I can't see any difference. Except, of course, faster access time and loadings, but it isn't make me as much happy as it should. Last culprits, in my opinion, are CPU, GPU and motherboard. If some of these will be damaged, I really hope that will be my processor (how strange it sounds!). It is easiest to replace and much cheaper than graphics card. But this my absolutely last hope - I don't want to surrender before I try every single method.


Also, one thing. If my crashes sound little similar to blue screens, maybe my PC is creating somewhere dumps or logs? If you know where I should look for them, I will post it here


Hello, it's me again.

After I used last method I found, I've been positively surprised because everything looked fine. I even tried to use my game-benchmark, Dirt Rally 2.0, and I spent in it couple of days without any crash - that never happen before. I was also so pleased, that I almost decided to post it in here as solution with many, many thanks for helping me. But...

looks like I need more help, because this **bleep** WHEA logger showed it's ugly face one more time. This time it was a little different. Happened in Snowrunner but not in game - in graphic settings. I did not changed anything, just only scrolling and reading descriptions. More than one month ago I had similar crash but I didn't paid attention to this little detail... and a fact, that I changed antialiasing to higher value (then instantly back to previous value). Maybe it tell you more and you help me to get rid of this :((

Also, I tried your method about changing frequency in Adrenalin and I am not sure I did it correctly. I clicked/checked everything like in your post but I have a little different interface with chart instead of bars. That's the screenshot:

In case, "częstotliwość" is "frequency" and "napięcie" means "voltage", so don't be scared my language ; ). I didn't changed anything except "P3" from 1750 to 1650. 100 less, like you told me


Well, your graph mostly corellates to maximum frequency, rather than minumum, but as it also allows you to edit voltage on same graph, it kind of more useful. But it is good to set stability

Quick switching through settings that cause immediate renderer update can cause sudden crash, technically. Like quickly swapping badly set OGL from windowed to full screen several times in row. It isn't consistent crash, but it can happen. 

You can try to set 1650 to 950, or 1750 (frequency you had before) to 950 voltage. It should increase stability on every step. But will hurt performance and heat a bit, as you will use higher voltage. If crashes became rare and erratic, it means you are close to borderline of stability. Even though question remains if it was GPU, why it became unstable in the first place? 


Okay, new settings looks that:

Freq 800 -> voltage 800
Freq 1275 -> voltage 800
Freq 1750 -> voltage 995

Why a little different? Because I accidentally changed value by moving red dot on graph so I had a tiny mess in values. Now I left default (but in Custom of course). I don't know why values became little different than before. Should I stick to 1750-995 or lower it to 1750-950?

Also, I have no idea why and what is unstable and I am asking myself the same question. I think that last 3 weeks were extremely lucky, or... or maybe, as you said, I am closer to achieve full stability. I hope the second one is true and I will change some settings/values with your help to make this crashes leave me forever


If 1750/995 is default, then i guess. 
Moving point to below = undervolt
Moving point to the left = underclock
Moving point to the right = overclock
Moving point to the upside = overvolt

You may want to find point on curve of relationship of frequency to voltage (what voltage core will take at given frequency), where system is stable. But you better not to go with voltage higher than default value, up to 25mv, so max steps are 1050 1025 1000 975 950 925 900, choses closest one you have to default value


So I should stick with values I typed above, right? And watch what will happen?


I guess. But you better reset them once and check defaults.
Depending on defaults it may be 1000 mV or 950 mV. And 50 mV is quite a decent difference. Usually between stable or not lies step of 6-18 mV
If unstable you can either increase voltage (but not too much, wouldn't recomment increasing it to more than +25-50 mV over default. Or decrease frequency, 25 mHz per step.


Hello DimkaTsv, it's me again. Sorry for writing from second account, but for some reason AMD Community don't recognize my password anymore and it's not sending me an e-mail to change it - but creating new one worked perfectly and I get mail instantly.

So... Maybe you and other users reading this thread thought that my problem is solved and I am just an ungrateful man that got rid of WHEA logger crashing without leaving a solution, but it's absolutely not like that. Several months have passed, when half of them was working (no time to play and cause crashes) and other half was struggling with it. You gave me some info to change clockings and voltage, so I've been trying different combinations with no effect. A huge revolution (as I thought) was unistalling normal drivers and downloading PRO version of them (it helped every other users with similar cause, wow!). First month was absolutely amazing, no crashes at all even in 3 weeks with Dirt Rally 2.0 - my "worst" game in terms of WHEA logger events. I intended to leave it here as a final solution but suddenly... yes, the black screen and terribly loud buzzing sound that made me almost jump on my chair. I tried to change values again, but my problem is not gone. Yes, PRO version of drivers improved my stability, but still it's not a full stability, so I am just scared about spending time in games... Not due to crash, but mostly about noise that it usually make...
I am so confused and desperate right now. Even reinstalling Win 10 didn't solved it. Do you maybe know any other ways to deal with it? It's making me crazy



a normal cpu / motherboard / memory etc should work with load optimize defaults and all set to auto and xmp enabled. If not it could be bad cpu with some tolerances perhaps too low. What have helped me in my 5800 problems was in the cpu PBO section to state it was motherboard in controll (set it for advance and then motherboard). Normally you should be fine with all set to auto. Did you update bios - a new AGESA has been released or something like that. A new bios update could fix things if you have problems.

Don't know what cpu / motherboard you have - but I know a lot of people have problems with the 5800 x cpu.