Processors

[RoughToughBanana]

Yes I know there are tonns of threads here with that problem and google have it allthough. But there is not really a answer to this  question?

My System:

Ryzen 9 5950x

MSI MAG x570S tomahawk

Corsair corsair vengeance rgb pro 2 x 16 GB 3600

Gigabyte Aorus 3080ti

Be quite Straight Power Platinum Pro 850 W

Bios is in stock mode nothing changed on settings and the bios is up to date! 1.20

Everytime after the computer was out for a time there are these WHEA messages:

Gemeldet von Komponente: Prozessorkern
Fehlerquelle: Machine Check Exception
Fehlertyp: Cache Hierarchy Error
Prozessor-APIC-ID: 2

So i found many many threads and infos about that but what exactly is it?? Some say CPU defect so RMA some say Board and Power issue?? So can someone help out plz!? Is the only thing in can do to RMA the CPU?

[Electric_Squall]

Disable PBO or return it (RMA)

It is a known AMD zen3 design flaw issue. 

Unfortunately, no response from AMD since 11/2020. They know it is a flaw and wont fix. AMD is dealing with it like these issues doesn't exist. The same behavior we see from this company, regarding to basically every problem their products have.

[RoughToughBanana]

Nice to Heat so what can I do? I have 14 days to cancle my buy 

[Stef64]

I don't work for AMD, I don't represent AMD, but what are you talking about?

Design flaw of what?

The CPU works just fine, no design flaws there, even Intel has these errors.

Why?

Did you troubleshoot? Do you have any idea what is going on beside RMA?

Get rid of the Nahimic garbage first. Update chipset drivers, video drivers and come back.

Took one full day of debugging to get to the bottom of this.

PBO2 is great, but one needs to know what they are doing once the system is stable and rid of bloatware.

The information is out there, you need to know what the heck you are doing.

My money is on AMD, sorry no spy chips from Intel, where security is compromised for great benchmarks.

[Aeitrium]

i just finished building my new pc last night spent nearly 12 hours with the whea logger events i could barely even install windows without it restarting, im pretty sure i got it figured out first thing i did was update bios through bios environment, its related to the speed of cores, voltage and silicon binning and maybe even power limits... i noticed a decrease in restarts as i lowered the maximum clock overide in pbo -(negative)250 for a total of 4.7 highest clock on any core and turned pbo limits to disabled (i have limits on manual now), i had tried undervolting it with optimizer curve before i messed with clocks and i kept getting whea problem.... so ive done some reading and im testing each core for 15 mins in occt physcial and virtual avx2 for 15 minutes 1 core at a time, i get whea errors at -19 on core 0 so i turned it to -17, -18 runs fine just want that gap for increased stability if problems arise. i think they had just pushed the limits of the cpu so high to maximize clocks and power out of the box. its possible its either trying to pull higher clocks than what the socket wants to give or the clocks are pushing further than what the chip can do. so it seems either at stock default bios u would have to increase voltage or the chip just cant handle the clocks, and id hate to increase voltage as its already pretty high  at those speeds, im running 4.5 all core and 4.7 single  1.34 volts transient, around 1.23-1.26 usually... if u can turn on pbo to advanced turn limits to disabled or default values in manual. and lower the maximum cpu clock overide abit it would prolly work just fine without messing with curve optimzer. i wouldnt rma as alot of what i read it fixed pplz problem but only for a time, jsut a few short weeks to months later there cpu is doing the same thing. and since it doesnt really freeze while under laod only mostly when next to idle i think its a voltage and clock issue cuz it pulls the most voltage when 1 core is clocked above the rest and the cores shift there clocks for SC so core doesnt create to much of a hotspot on cpu while the regular clocked clocked cpus... if anyone needs help i can guide them through abit of it....EDIT....((((TLDR)))) BEEN SCREWING AROUND WITH SYSTEM ALL DAY THE ACTUAL FIX IS TO DISABLE GLOBAL C STATE CONTROL IN CBP IN BIOS, IVE TURNED IT ON AND OFF  TESTING IT FOR HOURS, WITH IT OFF I DONT GET A CRASH AT ALL JUST SITS THERE FOR HOURS, THE MOMENT I TURN IT ON AND BOOT IT CRASHES ON WINDOWS STARTUP OR AT DESKTOP SOON AFTER, EVEN WORKS WITH STOCK SETTINGS.

[Shodan0000]

For sure I am going to test the above C-state solution. But currently I actually gave a positive number in curve optimizer on the core that gave me the Cache Hierarchy error. This seems to be stable.

Unfortunately no stability test program is able to generate the issue / error, only Red Dead Redemption 2 will crash and sometimes give the error. 

Will report back later when I had tried the C-state solution.