Showing results for 
Search instead for 
Did you mean: 


Adept II

Ryzen 5600x: System constantly crashing/restarting WHEA-Logger ID 18 and ID 19 sometimes

I'm having the exact same issue on a 5600x with a msi tomahawk b550. I have a x570 gigabyte aorus master and a r9 3900x I was thinking of swapping this CPU with that and testing there.

OK; Get ready for this, Here's everything I've done so far to eliminate variables:

  1. Swapped the system drive (nvme) and reinstalled windows from scratch
  2. Unplugged all other drives
  3. Swapped GPUs
  4. Swapped PSU
  5. Swapped memory from 3200Mhz Gskill TridentZ CL16 to 2667Mhz CL16 Corsair Vengeance
  6. Swapped the Msi Tomahawk B550 with another MSI tomahawk B550 to verify it wasn't the board.

Ensured I'm on the latest bios for the MOBO and the previous 1 which was also post Ryzen 5000 release.

7C91vA51(Beta version)
Release Date
Release Date
- Updated AMD AGESA ComboAm4v2PI Patch C
I have literally done everything you can do aside from swapping the CPU and turning off PBO or any other bios settings. So if anyone is wondering if this exact error and restarting behavior is a single component in the system, in my case, it's not. It's the CPU or the microcode that hasn't been updated yet. Now.... I've seen forums where people with 3000 series CPUs were getting the same thing. I'm at the point of buying a 5950x *(when it's available) to replace my 3900x, buy it's super annoying that I may have to RMA my CPU for my main gaming rig the 5600x when there is almost no stock and I got lucky enough to buy a CPU with the supply being so low.

For reference, these are the CPUs I've owned, none of which have had this issue:
fx 6300
fx 8150
r5 1600
r7 1700x
r7 1800x
r7 2700
ThreadRipper 2950x
r5 3600x
r9 3900x
Obviously, being a computer engineer and a HUGE AMD fan, this behavior is concerning to me. Especially when I'm patiently waiting for the 5950x and a 6900xt.
I'll put the tridentz back in and try disabling CPB And PBO.

GPUs tested r9 390x and rx 5700xt
1 Solution

For the record, I returned the 5600x and bought a 5900x.



Worked out of the box.




View solution in original post

16 Replies

Sorry to hear about your issues. 

While I too think stuff should work at launch correctly in reality that often doesn't happen.

This is why I tend to not early adopt on new hardware. I wait for a few months for bios, chipset driver and Windows updates to stabilize before jumping in. 

I can only tell you that from the way it collective looks from most the posts that this is likely something a bios update will need to fix. It is certainly possible you could have a bad CPU but likely it is just a firmware problem.

I would contact your motherboard's support department as they can best help diagnose and offer maybe bios changes or a new beta bios yet unreleased that could help?

You can also ask AMD support here:

If that doesn't help and you are in your retail return Window still you could return or exchange.

If you need to RMA it, you can start that process here:

Good Luck!

I should add that some find relief by lowering the infinity fabric speed and keeping the memory speed between 3200 and and 3600 if your memory supports those speeds to begin with. Don't over clock your memory to achieve that.

There are lots of other threads on this you can search and read with lots of advice that works for some and not others that may be worth a try.  

Yea; I mean I'm not doing anything other than the xmp profiles for my b550 tomahawk.

I've also tried with 2133Mhz as that's the default dd4 speed.

I mentioned two different sets of memory modules (installed in slots B2 and A2)

The memory modules I tested: G.SKILL TridentZ RGB Series 16GB (2 x 8GB) (PC4 25600) F4-3200C16D-16GTZR

Using the XMP profiles that are presented in the latest msi b550 tomahawk bios (I haven't tested on the gigabyte aorus x570 yet) Bios version is listed above. These profiles match the timing and frequency that my RAM is rated for.

NOTE: I used both an r5 3600x and a r9 3900x in this same exact setup with no issues.

  • Turning off PBO dramatically reduces the occurrence of the Blue screens
    • So far on random reboot in 7 hours of playing/testing
  • Turning off CPB brought my clock frequency to below the base frequency
    • this dramatic reduction in clock frequency is not acceptable for my use case


Obviously the expectation is that I shouldn't have to adjust a bunch of parameters for the processor to be stable and hit decent boost clocks that are under it's rated maximum boost.

My thought is that the following scenarios may be true:

  1. The silicon I received is not stable at the frequency that CPB boosts to (4.3-4.4 Ghz seems to be the case)
    1. This would indicate a sub-par or in the lower percentile of stable yield as it doesn't hit the frequencies expected without crashing the machine. This may be expected with initial yields being a slightly higher indice of more defective dies as it's a new architecture on this manufacturing process.
  2. A BIOS update may or may not fix this because it could be a microcode issue
  3. I may only be able to achieve "normal" functionality by adjusting variables and frequencies such as the infinity fabric or the fblk etc.


My concern here is whether or not I received a product that's inherently inferior to a vast majority of the dies produced.


The only thing I really care about is whether or not the chip itself is defective. If it's not defective and it's juts a firmware/software issue, then I'm perfectly fine with that. I deal with that all day long at my job. If performance or stability is being affected by a raid controller firmware or a nic firmware or something at that level, it's perfectly understandable and typically a firmware upgrade that contains the solution is the answer.


I just don't know how to differentiate my particular sample from a defective sample. I also don't know what the manufacturing QA looks like.


Just for the FYIs.


The EVent ID 18 is a WHEA-Logger Bus/Interconnect error


Event ID 19 is cache-hierarchy.



Also; I haven't tested how Linux handles this particular distribution. AMD pushes stuff out faster than their QA can push the bug definitions to Engineering. Wish I could help somehow. In any case, I assume there's some automation missing or that the resource allocation for this is more nimble due to the competitive market and profit motive.


In any case, many many users who don't know better can be turned off by this, hence the meme in the GPU market of AMD driver issues. I personally haven't seen it, but I'm not always playing the latest games, so I believe people when they say it.


Many of my engineering friends say they won't buy AMD cards because they somehow believe the drivers are inferior because of these early QA teething sessions.


I should add that I updated the bios for it's current board:

Gigabyte X570 Aorus Master

9‎.68 MB
  1. Update AMD AGESA ComboV2 for AMD Ryzen 5000 processors support


not change the behavior of the CPU at all. I reset the BIOS settings to default with the only exception of running the memory running on xmp profile at 3200 mhz and it's native timings (G.SKILL Trident Z RGB (For AMD) 16GB (2 x 8GB) 288-Pin DDR4 SDRAM DDR4 3600 (PC4 28800) Desktop Memory Model F4-3600C18D-16GTZRX)


I should also report, I rebuilt the machine on the x570 gigabyte aorus master and found that it made no difference. I'm convinced the CPU is the problem and that it's possible that the micocode to deal with whatever variation there is between most 5600x and mine doesn't handle the behavior of my sample.



I think I just go unlucky with an unstable sample. Not happy that supply is so garbage right now that a replacement could take weeks or months.



I have  same or very similar problem.

MSI b550 gaming edge wifi with 7C91v151 and 7C91v153 UEFI. F4-3200C14-32GVK.

Now it is my 4th unit of 5600x. Different batch numbers/series/shops.

First was nothing special but stable (low boost).

Second one was a good one ( frequently was boosting to 4.7 and more with PBO Enable as advanced + enabled motherboard limits + scalar x1 + 200MHz boost limit + curve negative 15 all cores)).

Third one was a bad one - unstable with PBO enable (+ motherboard limits + scalar x1 +200MHz boost limit + curve negative 15 all cores).

But the forth one...

The forth unit of 5600x is so so terrible bad...

It is unstable EVEN with PBO  completely disabled.

right now it is a nightmare with quality of 5600x cpu. 

Much, much better will be (high % of probability of a "good" unit)  to buy one of 5800x or 5900x.


For the record; I bought a 5900x and returned the 5600x. Issue was resolved.

I think there is a higher percentange of 5600x that are junky TBH.





For the record, I returned the 5600x and bought a 5900x.



Worked out of the box.





What software was used to reliably get BSOD?

I want to return the cpu but the retailer may say there is nothing wrong.

The cpu passes tests (prime 95 AIDA64) but BSOD in games after about 20 minutes. Even BSOD installing windows.

This problem still exists in mid 2022. I have a really bad 5600x, built another pc just to test the cpu and it is the problem. Like others turning off PBO helps, but using the default values and the latest bios the cpu should be 100% reliable, not a pain in the a..

I made the pc for a nephew and ended up giving him a new i7.


i have a 5900x and its been a nightmare since i got it its so unstable that it reatarts in the bios sometimes i have extra gpus for physx and other decoders and when i take them all out but 1 its a little more stable but still every 10 minutes and some times luckily maybe 4 hours it restarts and it doesn't matter i run it at 4.8ghz and it runs fine for a little them crashes even at 3ghz it crashes i think it is a bad dye because none of my coworkers have said anything about any issues and i cant find much online this thread has semi informed me im gonna try to replace my 5900x good thing i got the warranty at microcenter XD

Adept II

My problem was some faulty extension cables but also I noticed this motherboards have a double pin lock system and I did have trouble with that. Make sure your gpu is installed correctly, I know it sounds dumb but believe me I've build many PCs as well and this motherboard is tricky. The metal thing they added to the PCIe is tricky. It's suppose to fit in like a DDR4 double lock system. The right one is the regular "click" lock, the left one kinda clicks but not at the same time. Tricky, check that out bro. 


I rebuilt the machine with a brand new exact same motherboard. I swapped all components with other components that are known good and did it 1 at a time with 4-8 hours of testing between.

I literally put 2 different GPUs in it and I've been building PCs since 1998, I know the card is properly seated in the PCI bus and has in fact been reseated and swapped out various times.

Adept I

I have the same problem even replaced the CPU and it's still the same...Any other solution?



Did you replace with another 5600x?

I bought one last year, caused me a real pain as it was for a customer, passed testing sotware but bombed in games, sometimes it took an hour other times 10 minutes. Bought a 5500 for testing pc, no issues. 5600x have issues AMD wont admit it. They had been out for 2 years when I bought mine, still got a bad one.


This Ryzen 5 5600 of mine was working flawlessly for 2 years until one day my friends PC with Ryzen 7 5800x was having a restart issue so we swapped the processors and motherboards. His processor was still randomly restarting on my mobo . I switched back to my processor but now even my PC started to randomly restart. So I don't think the chip is there to blame cuz it worked fine for such a long period. My PC doesn't crashes in bios only except one or two times in 2 months.