What is wrong with the 14900k?

[vmovups]

Hello, i bought a 14900k with the accompanying motherboard, RAM and cooler in early November but have kept having problems with it.

The problems started with abnormally frequent crashes in applications powered by chromium that progressively get worse and random BSOD at rest but not with high usage, the distributor i bought it at replaced the RAM for another brand which did stop the BSOD but not the applications crash, after much insistance and finding out crashes stopped when the two 6ghz P-cores were disabled and the computer crashed programs and displays a BSOD very fast when only the two 6ghz P-cores were active they replaced the processor in early January which did indeed stop the crashes.

The crashes stopped for a few weeks until last week after a period of using the computer a lot the crashes returned, this time affecting games too.

When analyzing the crash dumps of applications through WinDbg, some anomalies which hint at the CPU being at fault can be found which i organized in the following list :

1. The CPU "ignores" the register content when using a memory operand such as [rcx+rsi*imm], in this example if RCX is a valid pointer and RSI is 0(implying access of the first element in an array) an access violation with address 0 occurs instead of reading [RCX]'s content.
2. The CPU triggers a write access violation when reading memory or executing an instruction which does not read or write to memory such as (SSE encoded)ucomiss xmm10, xmm0, the exception address can be garbage with typically all bytes except the lower one filled with 0xFF, another integer register or the contents of an XMM register which may be a floating point number.
3. The CPU is off by one when incrementing the RIP register, resulting in the CPU attempting to execute the last byte of the instruction it has executed or the second byte of the instruction it was supposed to execute.(see attached image, the CPU tried to execute the memory operand offset which was 0x60 as an instruction, but 0x60 is a reserved invalid instruction)
4. The RIP register no longer points to seemingly valid code.
5. Stack overruns exceptions or errors in which the call stack becomes unretrievable.

The first CPU only had 2 and maybe 1, the replacement CPU has developed all 5 since last week.

Application/windows hangs also occur on this replacement CPU, requiring manual restart of the application or windows session.

The replacement CPU may BSOD when a game's process affinity is set to E-cores only, the crashes ceased when the first CPU had P-cores 4 and 5(the two ones that can go to 6ghz) disabled through the motherboard's settings, the replacement CPU stops crashing when the power plan is set to power saving and still crashes when the 6ghz P-cores are disabled.

Both CPUs in their degraded state BSODed with ATTEMPTED_EXECUTE_OF_NOEXECUTE_MEMORY exceptionnaly fast(usually under 10 seconds) if only the two 6ghz P-cores and E-cores are enabled and execute an app which makes heavy use of AVX instructions on all cores, i am unable to reproduce the issue with benchmarks and a normal configuration as the CPU appears to be too power hungry to maintain its normal frequency with a ventirad and i am unwilling to use a watercooler.

Now that i explained the problem, my questions are how common is this problem exactly? what is causing these never seen before issues of CPUs degrading at an abnormal speed? Are the i5 and i7 affected too?(i need to know as the distributor i bought the hardware at might propose infinite replacements of the 14900k or exchange for i5 i7 or other brand) I see many other people complaining of more or less the same problem with their 14900k or 13900K(S) and i start to think the problem may be even more common considering how vicious it is, presenting as applications or games suddenly closing or a BSOD which may make the user blame the app developer or operating system itself.

heres my hardware configuration if it can help spot a pattern :

MSI Z790 Carbon

CORSAIR Vengeance DDR5 7200mhz

Intel i9 14900K

Dark rock Elite(Be quiet ventirad)

MSI RTX 4090

Be quiet DARK POWER 13

The previous motherboard that was in the computer never has these crashes so i don't think the GPU or PSU are involved, and disabling XMP does not seem to do anything, neither does the OS(tested with windows 10 and fresh 11 install), CPU frequency settings did not deviate from default.

Thanks in advance.

[AlHill]

Update the bios.

Doc (not an Intel employee or contractor)
[Maybe Windows 12 will be better]

[vmovups]

The BIOS was already updated with no change.

[Wesley1986]

I have a I9 14900KF and are experiencing the same problems. The only game that has issues and wont start is Battlefield 2042. 

I have the latest bios update as well and no change.

I did notice that the windows 11 scan for corrupted/missing files occasionally finds a few every now and then when issues start to arise but come back after a few weeks. this does fix the problem occasionally but not always.

I know my cooler is a 240mm and struggles to keep it cool but manages 40 degrees idle and 70-80 degrees heavy loads.

All of a sudden now my discord quits randomly and cant open it back up until a restart but happens a bunch.

ASUS H670 TUF GAMING PRO WIFI D4

32GB CORSAIR VENGEANCE DDR4 3200MHZ

INTEL I9 14900KF

MSI RTX 3070 TI

CORSAIR 850 PS

COOLERMASTER MASTERLIQUID ML240L V2 RGB

[vmovups]

Can you send discord or BF2042 crashdumps? They should be in %appdata%\discord\Crashpad\attachments and %localappdata%\CrashDumps

(some games put their crash dumps in a special folder, else do a search on C: with *.dmp)

[Mj3mej]

Exactly same problem

Rog z790e wifi II
Strix 4090
Corsair 32 ddr5
Nzxt 10 fans 3 watercooler

[vmovups]

Can you send crash dumps? They should be in %localappdata%\CrashDumps , if they're not there searching C: with *.dmp should list all crash dumps present on the hard drive.

[SHAB]

I see this happening with new gen and older gen.

Try this solution:

I'm new to this forum but i been having the same problems with my 13900KS my only solution to get system stability is to bring down the P cores ratio down a notch. Going from 56x-60x factory setting bring it down to 55 or 54x. I'm going to eventually RMA this chip.  You can do this via the BIOS setting depending on your motherboard. Read this post as this was relevant to my issue. Try disabling turbo boost in bios setting and see if you have system stability at 3GHZ not ideal and loss of alot fps, but it can rule out that the cpu could be the problem. The solution above is keeping everything enabled at bios default except changing the P-core ratio value.

Is there a fix for the 13900K crashing with Intel Turbo Boost on the Windows 22H2 Update? : intel (reddit.com)

Edit: Also been having BSOD and numerous issues for the past year because of this and have come across this thread in Reddit. All crashes Gone and no BSOD and no tabs crashing.

[vmovups]

It's not normal, i already RMAed the chip(which is difficult as is because it doesn't occur in benchmarks and the distributor wants to exclude every potential cause) and the replacement CPU has become worse than the previous one! The original CPU had zero crashes if the 6ghz P-cores were disabled, this one crashed the game i wanted to play every hour on average.

What i mean is don't expect RMA to change anything. For me the new chip started crashing software again after three weeks i believe and has become worse over the week.

[SHAB]

All you need to do is verify if maybe the factory overclock is unstable. It happened to my chip 13900KS, maybe yours is also overclocking and unstable. Set your P-Core Ratios at 55 or 54 and see if the game doesn't crash. IF doesn't then you will know the overclock is unstable.

[vmovups]

The chip is by definition not overclocked... It quit the factory having presumably passed tests and i do not believe transport could have caused invisible damage that would present itself as nonsensical errors...

Either the factory tests are too loose or the motherboard defaults(which i trust match intel recommendation) are destroying the CPUs. Possibly both since there are both cases of degradation and cases of the CPU appearing to be in this state from day one.

That said you are right that reducing P-core ratio appears to stop the crashes, but we paid a lot for a CPU that apparently needs to be RMAed every few weeks, i start to think Intel is perhaps hiding something.

[SHAB]

Seems to me that Yes possibility Intel is hiding something, but also I'm not going to fret over a few 100 MHz speed from 5.8 to 5.5 GHZ. Overall i think this solution is the correct one for alot of us that have these odd problems because the reddit thread is archived and alot of people that have these problems came from the windows update from last year that changed something.

I went through all of these problems by switching components and such but never realized that the CPU itself was the problem. A few Bios Tweaks and it stabilized. I rather have stability and use my computer than fret over a few 100 MHz speed.

[vmovups]

Still no official answer from Intel?

[Ricbun]

Exactly the same problem here. I have some of my custom made applications crash in which I can see the exact crash and causing exception and I see a lot of memory access violation and corrupted memory related exceptions. Commenting to see if Intel replies..

[SHAB]

Try this solution and see if it works for you, Intel support is very lackluster.

Try this solution:

I'm new to this forum but i been having the same problems with my 13900KS my only solution to get system stability is to bring down the P cores ratio down a notch. Going from 56x-60x factory setting bring it down to 55 or 54x. I'm going to eventually RMA this chip.  You can do this via the BIOS setting depending on your motherboard. Read this post as this was relevant to my issue. Try disabling turbo boost in bios setting and see if you have system stability at 3GHZ not ideal and loss of alot fps, but it can rule out that the cpu could be the problem. The solution above is keeping everything enabled at bios default except changing the P-core ratio value.

Is there a fix for the 13900K crashing with Intel Turbo Boost on the Windows 22H2 Update? : intel (reddit.com)

[Ricbun]

Thanks! Tested this by setting to 54x and already seems way more stable

[Christoph-Huber]

Same problem here,

CAD-Machines, ArchiCad, TwinMotion, Enscape, DX12

The 3D is not stable, no system crashes but application "ghosts" on about 20 machines.

I9-14900KF
ASUS PRIME Z790-P
Kingston KVR56U46BD8-32 5600 MHz
NVIDIA RTX A2000 / TRX A4000
Samsung SSD 990 PRO M.2 2280 NVMe
Win 11 Pro, 23H2

To me it sound like the gaming problems. Same tasks only different graphic card.
Same operations on older HardWare with the same data does not show any problems.
Like i9 9900 .... 

Any explanations from Intel?

[Christoph-Huber]

Actual Research and state.

"Stress tests" do not crash the machine. But ArchiCad 26 and 24 has reproduceable bugs. These are now reviewed by Graphisoft.
With every ArchiCad Crash I have some Sys-Log Warnings, WHAE-Logger Event 19, "Recovered HW Error", Translation Lookaside Buffer Error, CPU APIC-ID:40 or ID:48

Tested with latest (Feb 26) Bios and Drivers from ASUS for Z790-P on a faulty machine.
But I also found identical machines that do not have the symptom.

Through my Assembler the case has now escalated to intel and maybe will also to Asus.

Since Game programmers blame intel for a similar problem, I also expect Graphisoft to end at the same conclusion.

So, does Intel red here and what’s the statement of Intel to these problems?

[MTcore]

It so often is Asus mobos I see when reading the 14900 crash threads, do they just sell that many more boards for 14900 than other brands?

I had these issues with Asus z790 tuf Gaming pro wifi and 14900k,

Main problems: 'Defaults' are not safe especially after enabling something like XMP.

1) SA voltage was ramped up on my system as a 'default' causing hangs everywhere (in UEFI and OS) without any obvious message to say, hey we at Asus thought going from 0.8v to 1.35v would be no problem and if you complain to us we'll blame the vendor approved list and YOU. (They pissed me off)..
2) 'Trained' SVID tends to heavily under-volt and for me, this meant many or most tests pass and then random crashed occur during general use, bad read type BSOD but far more common, app just closes. When testing, the use of Intel Thermal pref boost or what ever it is masks the issues.. it holds the clocks low due to temp in the test but allows games to bink the higher clocks as the temps are lower finding the crash points.

Fixes,
Set SAv to static at a safe voltage.
Where the bios has undervolted too far add +voltage offset and/or adjust the LLc depending on how bad the issue is.
Actual fix would be re-educate the mobo manufactures.

My system:
Asus z790 tuf gaming pro wifi
2x 32mb Corsair Vengence DDR5 @6400mts
i9 14900k, no power limits, heavy (all core) upto 5.9, light (on four cores) upto 6.2 (E cores 4.4-4.6)

Palit Darkrock 4090
Deepcool 850w psu (yes its more than fine, believe me I've tried to make it fail)
Artic LF 3 420 (omg what a beast).

FYI, While experimenting with the advice in these forums I tried 'Intel's fail safe' saw my chip suck back 475w o.O and the cooler just went... ok.. 95c you shall be ;D might wanna watch out for that on lesser coolers..
PS I did also lap my CPU... the 'Mountains, Valleys and Canyons' style IHS Intel provides was holding me way back..

[Colosio]

i buyed 14900k for fornite gaming, and it ends up fornite being one of the MOST afected games by this procesor, it crashes and gives out of memory errors and its just because of the 14900k....

[Daan0]

This has to do with power limits of the cpu, some mobo manufacturers set it way above the designed specs.

It took me 3 days before i figured this out.

Change your cooling profile from water cooled to air tower cooled.

This artikel solved it for me

https://www.tomshardware.com/pc-components/cpus/is-your-intel-core-i9-13900k-crashing-in-games-your-motherboard-bios-settings-may-be-to-blame-other-high-end-intel-cpus-also-affected

Hope it helps.