cancel
Showing results for 
Search instead for 
Did you mean: 

Processors

CrispyCrunch
Adept II

Ryzen 5900x: System constantly crashing/restarting WHEA-Logger ID 18 and critical error Kernel-Power

Mainboard: MSI x570 Unify
Mainboard-BIOS: 7C35vA82 (Beta version)
CPU: Ryzen 5900x
RAM: Crucial Ballistix BL2K32G36C16U4B 3600 MHz, 64GB (32GB x2)
Drive: M.2 Samsung 970 Evo+ 1TB SSD
Graphics: SAPPHIRE Nitro+ Radeon RX 5700 XT
PSU: be quiet straight power 11 750w Platinum
OS: Win 10 Pro (64bit) - all updates installed
Chipset driver: 2.9.28.509 (released 2020-11-09)

I first assembled the PC with a Ryzen 3800x a week ago because it was unclear if and when I would get the Ryzen 5900x I ordered. Worked with the included AMD Prism Wrath CPU cooler for one week without any problems.

- Today I installed a Ryzen 5900x and a Scythe Fuma 2 CPU cooler.
- After 20 min the first crash/restart with the following entries in the Event Viewer: WHEA-Logger ID 18 and critical error Kernel-Power ID 41.
- Happens irregularly again and again, sometimes after minutes, sometimes longer: Windows freezes for a few seconds and then the PC reboots. Doesn't matter if load or not.
- CPU temperature between 30 and 40 °C
- Updated to BIOS and chipset driver mentioned above: Problem still exists
- XMP Profile disabled (RAM on 2600 MHz): problem still exists
- CMOS Reset: Problem still exists

Either there is a compatibility problem of something with the CPU, or the CPU is defective?
What to do? Really frustrating.

2 Solutions

Im having a similar issue, x570 aorus and 5600x. Have same errors on windows. 

Disable CBP and PBO and run it at default settings (3.7 ghz and xmp on). That works for me. 

View solution in original post

I got a new angle on this. So deactivating PBO and CBS definetely works, PC was running stable for a week now. But you'll loose performance.

So I wrote to the MSI support and the AMD support.

MSI suggested to try increasing the DRAM Voltage by 0.05 V, which I did. System seems to be stable, no crashes so far - neither in idle or while gaming.

View solution in original post

947 Replies

ASUS B550 Gaming-F with BIOS 1401 (was removed today from ASUS page??)

G. Skill Trident Z Neo 3600 2x16 GB CL16

5900x

WHEA error and restart every time I close game (PUBG). Deactivating PBO and CPB in BIOS is the only way to avoid this BSODs.


@Neutra is your restart also relatet to closing PUBG?

Should I RMA it?

 


@Lenny29 wrote:

ASUS B550 Gaming-F with BIOS 1401 (was removed today from ASUS page??)

G. Skill Trident Z Neo 3600 2x16 GB CL16

5900x

WHEA error and restart every time I close game (PUBG). Deactivating PBO and CPB in BIOS is the only way to avoid this BSODs.


@Neutrais your restart also relatet to closing PUBG?

Should I RMA it?

 



Are tou running any RGB control programs? Did you updated bios with the internet option?

0 Likes

I had no problems with my new rig with a 5800x on a msi b550 gaming plus until i added extra RAM and went from 16 gb to 32 gb (all Corsair vengence 3600 cl 18) and ran into the problem with random restarts with kernel power reported in the windows log both under load and idling with standard bios-settings and with and without xmp enabled.

 

The solution for me (at least for now) was not to do any changes to the cpu but to set a fixed nb/soc voltage to 1.11 and the ram-voltage fixed to 1.44.

 

Regards

0 Likes

Where did you get voltages from?

Its from trying and tesing?

Did you try for example something lower like 1.1 soc and 1.4 memory?

0 Likes

I was/am testing. With 16 gb i had no issues. With 32 gb and without any changes to the soc and ram-voltage (they were at auto) i had at least 1 restart every hour. With 1.11/1.40 i booted and could run cinebench r23 and idle for long times but PUBG (the game) kept craching. Though- 1.11 soc and 1.44 ram-voltage has not given me any restarts or WHEA-errors yet (2 hours gaming and about 12 hours idle). I should also add that i have 4 RAM-sticks and from other forums i have learned (after i bought the gear ofc) that ryzen often runs a lot better without having to experiment with ram-voltages if you run the cpu with only 2 sticks. Hope this could help someone. 

My theory is we lost the silicon lottery and, somewhere along the vcore-frequency curve Performance Boost calculates, there is instability. This is why (most) of us can run sustained loads fine, but we crash with a variable load like gaming. This could probably be fixed with an eventual Bios update that adjusts the curve, but we are all entitled to an RMA because the silicon we received does not work at stock speeds (and disabling PB, a stock feature, is not a solution).

 

I was able to keep PB on by setting my Vcore to 1.35V and LLC to max. My CPU now boosts up to 4.9Ghz stable at 1.35V (stock was 4.8Ghz at 1.5V), but it now idles at 3.8Ghz 1.35V (stock was 0.95V). So, I'm getting slightly higher than advertised speeds at a lower voltage under load, but am running slightly hotter at idle - about 40c vs 38c. 

 

I never planned on OCing this CPU, but now I'm curious if our silicon will underperform at the top end since this does seem to be an issue that presents itself outside of max load.

Actually I think this theory is right. Actually I think the instability is where it should be i.e. at the highest frequences which are achieved at light or variable loads, it's rather strange you stabilized the CPU by undervolting it at max frequences.

But I think what you (and other guys above) do is rather harsh on the CPU compared to the Curve optimizer solution. Curve +8 is just +0.05V at the maximum, and it's applied "intellectually" and not all the time. 

Also, in my case I was able to apply +6 Curve just to the 4 "worst" cores of the 2nd chiplet of my 5900x (you can check the cores order in CPU-Z) and get it stable. It's probably possible to stabilize the CPU with the + Curve applied just to 1-2 cores, need futher testing.

Also you probably should raise the EDC current limit to like 200A, it affected the stability in my case (as a supplemental measure, the Curve is the main thing). But I found out that doing it and leaving other manual parameters there at 0 cripples the performance as it sets the TDC too low. So you should either set the limits not to Manual but to Motherboard (with AT LEAST a Noctua D15 as your cooler, it'll do the job but raize the TDP limit to like 165W) or manually set the TDC at like 160.

But of course if you can RMA such a CPU with little or no pain you should absolutely do it.

PS: Actually looks like there are two sorts of problems here. The solution for the first type is described above, the second type is related to the memory controller and solved by raising the SOC voltage and probably memory voltage.

Anyway (for other new victims reading this) you should start with trying to disable the Core Performance Boost and turning off the memory XMP (but set the memory voltage at 1.35 if this is stock voltage of your memory).  If this helps you have one of the problems described in this thread, if not then you have some other problem which might not be even CPU related.

 

I decided to return my copy of 5900x

Let's be honest. Some ryzens are working, ours not.

If silicon lottery exists we won crap.

Decided to check prices and availability in next week or two. I think I'll go with 3900x.

I woukd like to buy intel but sadly, I can't return my motherboard now

Well, some others work too but with a little tweak and loosing like 2-3% of the performance. It's not normal but it's no reason to ruin the system and waste lots of time and/or money.

Mine is like this at the moment: https://valid.x86.fr/hrf3sd

That's why I said you should RMA only if it bears little pain. Going with a 3900x doesn't look like little pain.

0 Likes

For me rma never work.

9/10 cases I had to wait abiut 3 weeks onky to get an info that service can't find a problem and they have to return "device" eithiut repair. Then next rma, and If  lucky ...

This is my pc/audio/video 15 years kf experience

My cpu is currently on the way.

Instead of waiting a month for the answer from service I prefer to hunt for the next cpu.

If 5900x will be back in shops maybe I'll try once more.

Maybe 5800x price per core will drop a little...

If I get a discount for 3900x will go with this model.

Many possibilities

0 Likes

Ekran Alıntısı.JPGEkran Alıntısı1.JPGEkran Alıntısı1.JPG
when i check QVL list on Asus Page
it shows
19-22-22-42 timing

but when i check it with CPU-z it shows
18-22-22-42

So when Asus write new Bios, they make a wrong timing? is it possible?
 
have ASus tuf X570 mobo (AMD 3900x)

@860lacov I am familliar with most bios settings and don't mind getting my hands dirty testing stuff (lord knows I've tried enough stuff already) - but I am pretty new to Ryzen, can you explain what LLC is, and how it is applicable?

Most of my crashes have been at low load (web browsing, youtube watching, etc). Gaming/synthetic tests have been more or less stable. 

I'm not going to neuter my 5900x, but messing with undervolting/all core overclocks is something I'm willing to play around with. Wondering how LLC fits in to that though. 

Thanks!

0 Likes

Corsair SAYS (offical support)

 

Usually, the BIOS would not be public and you will have to wait for ASUS to reply.

Sure, that is an option ( about others drammanuf Gskill- Adata etc) but remember in general AGESA is causing compatibility issues for kits above 3200MHz so switching brands might not be the right solution.

My corsair 2*8 16 3600 CL 18 makes RANDOM RESTART only DOCP/XMP enable mode
they said wait ASUS bios. İ think corsair and Asus knows AGESA issues.
in 1 year ASUS realed 20+ bios. WHAT THE HELL, somethings wrong.

0 Likes

I ll try to make a long story short. (5950x, dark hero)
-Had the system for a week on an old windows installation, fine for that period of time
-decided to make a fresh install on Sunday
-started getting multiple WHEA reboots, almost all of them at idle (or simple use like browsing etc), only once during gaming I think
-yesterday flashed to the latest beta bios and disabled c-states only, (had DOCP enabled) etc
-left the pc continuous working since (~24h) and most of the time has been idle with the occassional game last night.
-no WHEA errors since

0 Likes

*accidental double reply

0 Likes

Did you jump straight for disabling c-states or try to run it 'stock' + xmp first? 

I have downloaded the beta 3201 for my ROG E-Gaming but not had chance to flash it yet.

0 Likes

I just wanted to reply with my experience (because I had been having issues with an x570i and 5900x for the last 6 weeks - random restarts and BSOD). 

I updated to Asus 3201 (AGESA 1.1.9.0) yesterday, and although the sample size has been small - it's been flawless so far. 

My ram is OCd to 3800C16 with FCLK @ 1900 (3600C16 is DOCP), I passed OCCT stress test, as well as Testmem5 w/ anta777 extreme preset. It's also been over 24 hours now with no restarts, BSOD, or WHEA errors in event viewer.

This may not be the solution for everyone - but for my issues it seems to have solved them. So keep an eye out for new bios's with the new AGESA patches. 

0 Likes


@willeywilson wrote:

Did you jump straight for disabling c-states or try to run it 'stock' + xmp first? 

I have downloaded the beta 3201 for my ROG E-Gaming but not had chance to flash it yet.


I *think* I ran it full stock for a few moments on the latest STABLE bios but then went straight to the beta bios + disabled c-states + DOCP/XMP. Which means I am not sure if the stability is from the beta bios or the disabled c-states. (who knows, maybe both), I'll give it a couple of days to probably verify that the WHEA are completely absent and didn't just become sporadic and I'll probably then while on the same bios enable the c-states and see how it goes. It should help differentiate which of the two stopped the errors. (obv fingers crossed is the bios)

Quick update from me (still have my problematic 5950x) being running semi-stable for a week or so at +2 ccx0 and +6 ccx1, with xmp 3600 and PBO turned on to MoBo settings.

Asus released Beta Bios 3201 today for my x570 e-gaming so I gave it a shot. I've got everything set to as it was before (XMP, PBO etc.) Except curve optimiser is left off.  Only had 10 minutes to test but it seems stable.  I could previously force a crash spamming cinibench r23 single/multi processes. This bios is based on AGESA 1.1.9.0. 

Only downside is peak processor speeds seem 50mhz worse overall, cinibench scores are worse and the temps are higher (actually hit 87°c on multiprocess benchmark, previously it would touch 83°c).  I'll do some more testing tomorrow. Fingers crossed.

0 Likes

RMA UPDATE: I received my replacement 5900X from AMD yesterday and it looks to be stable. The difference is like night and day. Only one day of testing may be too early to tell, but it's a big improvement from the old chip which crashed right after I first installed it. Below is a summary of everything.

SYMPTOMS: Received 5900X beginning of December that within minutes of installing for the first time had bluescreen WHEA crashes, usually cache hierarchy error. Normally crashed at idle or doing a light task like opening an app or web browsing, most often within a minute of Windows loaded. Sometimes was stable long enough to do a benchmark, but never got beyond 5 minutes after Windows loaded before crashing. Does not crash in BIOS. Chip date code: BG 2043PGS.

THINGS I TRIED:

All BIOS options set to default (no overclock)
Two different sets of RAM (G.SKILL TridentZ RGB Series, G.SKILL Ripjaws V Series)
XMP off (2133Mhz)
XMP 3000, 3200, etc
FCLK 1500, 1600, etc
All compatible BIOS versions (ASRock B550 Phantom ITX). Latest AGESA I could test: 1.1.0.0 Patch C.

WORK AROUND: I could make my system stable by either turning off core performance boost (CPB), setting power profile to "eco", or setting all core magnitude in curve optimizer to +8 (overvolting). I ran +8 stable for two weeks without any crashes. I preferred this work around because it did not affect performance as much as turning off CPB or using eco mode.

RMA PROCESS WITH AMD (total 4 weeks):

Dec 1: Submitted RMA to AMD
Dec 14 Received response asking for proof of ownership and purchase. I provided same day.
Dec 15: Accepted for return
Dec 16: I shipped via AMD's provided FedEx ground shipping label
Dec 21: AMD received
Dec 22: Approved for replacement
Dec 28: Received replacement

RMA RESULT: My replacement 5900X works great. It is like night and day. At default settings, I am able to set FCLK to 1867Mhz and memory to 3733Mhz, and undervolt CPU to -15 in curve optimizer. Previously, I had to overvolt CPU +8 to make it stable. I am seeing 5-10% better performance in benchmarks and at lower voltage. And this is running AGESA 1.1.0.0 Patch C, which is still immature for this chip and more can be expected from it with AGESA 1.1.8.0 and later.

@thunk_stuff  What is the BG Grouping # for the Replaced CPU? 

 

 

@jpee80 The bad chip was BG 2043PGS.

Unfortunately I didn't write down the BG Grouping for new chip before I installed it, and it looks like you can't get this code from the box.

0 Likes

Congrats! Hope it works now!

You gave me hope!

Just started the RMA Process.

BG: 2043SUS

0 Likes


@tim716 wrote:

Actually I think this theory is right. Actually I think the instability is where it should be i.e. at the highest frequences which are achieved at light or variable loads, it's rather strange you stabilized the CPU by undervolting it at max frequences.


 The thing is, during those variable loads, my CPU would bounce between full boost and near-idle states. One moment it would be 1.5V 4.8Ghz and the next 1.0V 2.7Ghz. Either there's instability somewhere along that curve or the Vcore can't keep up. Turning on LLC was the easiest solution.

 

I'll take a look at the curve optimizer. This is my first Ryzen CPU and first time manually changing CPU settings in 10 years, so there's some catching up with tech to do.

0 Likes


@jaydeejoy wrote:

TUF GAMING X570-PLUS BIOS 3201
Update AMD AM4 AGESA V2 PI 1.1.9.0.


Testing new bios over 2 hours


Asus Tuf X570


3900x


XMP on 3600 Cl18    2*8 16 Gb 
PBO on 1x

No Restart 

i hope new bios fixed that random restart Kernel 41 issue.
i ll let u know if i got restart again.

Ekran Alıntısı.JPG



Random restart again with new bios (TUF GAMING X570-PLUS BIOS 3201)
XMP/On
PBO/1x

 


now disable pbo
XMP/ON
testing again

0 Likes


@jaydeejoy wrote:

@jaydeejoy wrote:

TUF GAMING X570-PLUS BIOS 3201
Update AMD AM4 AGESA V2 PI 1.1.9.0.


Testing new bios over 2 hours


Asus Tuf X570


3900x


XMP on 3600 Cl18    2*8 16 Gb 
PBO on 1x

No Restart 

i hope new bios fixed that random restart Kernel 41 issue.
i ll let u know if i got restart again.

Ekran Alıntısı.JPG



Random restart again with new bios (TUF GAMING X570-PLUS BIOS 3201)
XMP/On
PBO/1x

 


now disable pbo
XMP/ON
testing again


RAndom restart again
in this settings
XMP/ON
Bios /Default

Now just

XMP /OFF 
Bios Default


0 Likes

ASUS removed their LASt BETA Bios(3201) on offical Page

 

Ekran Alıntısı.JPG

0 Likes

no information from amd or mobo manufacturers

I'm starting to think that it could be more serious problem.

 

Im also gettin WHEA 18 and "solved" the crashing by disabling PBO CBP.
Seems stupid to fix something by disabling features. Whats the impact on fps for games when these are disabled?

Try disabling only the CPB only and see if that works. (That turns your 5900X into a 3600 or lower)

I fixed my problem with a new ryzen 5900X.
I also did a complete wipe of all drives, reinstalled bios with USB pendrive option and installed windows 10 again without armoury crate from Asus.

Also, DO NOT install any RGB Lightning control program.

 

0 Likes


@Bvzz wrote:

 

Also, DO NOT install any RGB Lightning control program.

 


Why? (Serious question)

 

I managed to get my 5950x 'stable' with a +6 curve optimiser on ccx1 and +2 in ccx0.  Unfortunately I have multiple RGB softwares due to the case, the fans, the MoBo etc but they don't seem to bother the system at all. 

In fact, I couldn't even install windows without the curve tweak...

0 Likes


@willeywilson wrote:

@Bvzz wrote:

 

Also, DO NOT install any RGB Lightning control program.

 


Why? (Serious question)

 

I managed to get my 5950x 'stable' with a +6 curve optimiser on ccx1 and +2 in ccx0.  Unfortunately I have multiple RGB softwares due to the case, the fans, the MoBo etc but they don't seem to bother the system at all. 

In fact, I couldn't even install windows without the curve tweak...


Because I’ve read in many places this kind of programs cause high instability in some systems and make frequencies fluctuate even more. I was not believing it but after installing 3 programs of this type all my WHEA started.

Try to RMA the CPU if you can and get a 300 series in Amazon meanwhile.

I received a new 5900X and works 100% without problems.


@Bvzz wrote:

Try disabling only the CPB only and see if that works. (That turns your 5900X into a 3600 or lower)

I fixed my problem with a new ryzen 5900X.
I also did a complete wipe of all drives, reinstalled bios with USB pendrive option and installed windows 10 again without armoury crate from Asus.

Also, DO NOT install any RGB Lightning control program.

 


I tried to disable CPB only (PBO on auto) and this also works. My system will crash if I close the game PUBG and CPB is activated. No RGB software installed. But I tried out ASUS aurora software and this seemed like bloatware to me. Couldn't uninstall it completely and had to do a clean install of windows.


Still asking myself if a bios update will fix the problem or if it is just a broken CPU thing.

0 Likes

I checked and with multipler fixed on 43 and vcore on 1.3V systemm seems stable.

Checking lover voltages now.

If CPU with all cores set to 43 is stable then I'm starting to believe that it's really problem with poor bios optimisation

0 Likes

Also using the ASUS ROG STRIX X570 E-Gaming here with BIOS 3001, Trident Z Neo 2 x 16GB 3600 (with XMP enabled).

Random restarts every single time (with and without XMP) until I used the curve optimizer to +2 chiplet 1, +6 chiplet 2.  But even then chiplet 2 doesn't get anywhere near target boost speeds, I've requested a return and hope for a better product next go around. 

Whilst this may be a BIOS issue, there is no denying some of these CPUs aren't entirely 100%

 

@willeywilson wrote:

Also using the ASUS ROG STRIX X570 E-Gaming here with BIOS 3001, Trident Z Neo 2 x 16GB 3600 (with XMP enabled).

Random restarts every single time (with and without XMP) until I used the curve optimizer to +2 chiplet 1, +6 chiplet 2.  But even then chiplet 2 doesn't get anywhere near target boost speeds, I've requested a return and hope for a better product next go around. 

Whilst this may be a BIOS issue, there is no denying some of these CPUs aren't entirely 100%


I'm in the process of returning my 5800x. Tried in two x570's... MSI Tomahawk and Gigabyte Aorus Master, both crashing and rebooting. Mine goes under any sort of load, so can't even load anything more than an internet browser. Managed once in three days to load up Call Of Duty. Think that may have been luck more than anything. What to do... Try another 5800x?

0 Likes


@usurper wrote:

@usurper wrote:

@mla_xxiv wrote:

I have my R7 3700X running smooth with no issues before upgrade to R9 5900X that's when things get messy lol

Upon changing my CPU, i got this random restart on idle or casual browsing. I did try replacing it again with my 3700X and no issues whatsoever and swapped back the 5900X then the issue is still there. 

Tried turning off the PBO, XMP, updating to latest 3 bios of MSI X570 Carbon Pro still i got these random restarts. Only thing that works to me is i manually overclocked my 5900X to 4.3ghz @ 1.325v. (PBO-disabled, XMP-ON). It's been 3 days since i did it, so far i haven't experience any crashes or random restarts.


that sounds like my issue described here
https://community.amd.com/t5/processors/new-ryzen-3900x-x570-random-restarts-whea-logger-error-id17-... 

as an update, it had ran fine for a week, and since then i re-enabled XMP and it ran fine for 2 more days


ok, update #2 - after 3 days of doing fine with XMP back on, it happened again, almost right after boot, when literally nothing was going on and barely any apps loaded. so i'm re-disabling XMP...


i have exactly same issue

WHEN DOCP/XMP enable (2*8)C orsair 3600. tested 2 diffrents corsair kit and same issue.
with 3900x ramdom restart almost right after boot in low load. 
x570 tuf (lastest bios 3001 lasstest chipset windows etc)
when i DOCP/XMP disable , theres no any issue no restart .
mailed aSUS About bios issue.

0 Likes

Have the same issue on 5600x

Disabled CPB and manually set CPU clock ratio to for example 44.00 and V-core voltage to 1.250 and it works fine no crashes.

You can also try vcore 47.00 with volt 1.33

Did a Prime 95 stress test.

But now CPU will idle on fixed frequency,should I go for RMA?

0 Likes

update to my own problems wih the 5900x and kernel power id 41 ( 63 ) error code, after the computer turn off in games after 5-15-30 minutes.  ( defective cpu )

I have today bought a new 3600xt cpu and replaced the 5900x cpu.
Instant my pc run smooth and 0 crashes in games.

Very long story short i replaced all my hardware and did stress tests for 8 houres and even torture tests of the hardware.. with the 5900x the stress tests all passed and computer run stable all night doing tests.. But if turned on any game the computer would turn itself off after a few minutes.
The only way i could play games with the 5900x cpu was if i manualy went into advanced mode in bios and changed cpuVcore from auto to 1,375v then i could play for houres.. BUT this is not a good solution since the cpu will not perform as expected when it cant control the power itself.
I also notised that at auto or manual power settings the cpu no matter what would always run faster than expected out of the box.. without overlocking it would boost to 4952mhz on all cores now and then..all thoug hthat is great, i do wonder if its normal since the cpu specifications say it should run from 3700-4800mhz.. I did not overclock anything.
No matter what any VR game would crash the pc after 10-15 minutes even if i manualy control the cpuVcore.
My problem was simply that if i select factory default/load default settings in bios then the computer would crash in games.
That is NOT how its supposed to work, a cpu should work out of the box at factory settings and be stable.!!!

My conclution after 3 month of issues and instability is my cpu is defective and i have desided to RMA the cpu though its very difficult to get a new one.

My advice try the above, or be prepared to RMA your cpu and get a new one.. You properbly have a defective cpu.
ps.. pc magazines and talk on the net indicated that the new ryzen cpus have a higher than normal failure rate.
My cpu is one of the first that was avaliable to the public. ( they where sold out in less than 8 minutes )
I bought it at www.komplett.dk

If anyone want to know then these are the numbers written on my cpu / production number ect.

Ryzen 9 5900x
100-00000006   ( maybe there is a 2 after 6  not sure )
BG 2046SUS
4520117W00422
Diffused in usa
Diffused in Taiwan
Made in China
2020 AMD

Can can also try

* WITH CPB ENABLED

*Change Dynamic V-Core DVID from AUTO to NORMAL and offset it to +0.100 volt 

It should work for those who are getting crashes after Enabling CPB. 

0 Likes