Regarding Intel i9-13900K and 14900K stability/crashing issues

Published 2024-07-24
going over potential fixes for these high end chips, mostly applies to asus board owners

NOTICE: I AM IN NO WAY SHAPE OR FORM A PROFESSIONAL IN THIS FIELD, I WILL NOT BE HELD LIABLE FOR POTENTIAL DAMAGES OR DEAD CHIPS, FOLLOW AT YOUR OWN CAUTION

pc specs:
Intel Core i9-14900K
ARCTIC Liquid Freezer III (420)
Asus ROG MAXIMUS Z790 DARK HERO ATX WIFI 7
G.Skill Trident Z5 Neo RGB 64 GB (DDR5-6000 CL30)
WD Black SN850X 4 TB
Asus TUF OC GeForce RTX 4090
Lian Li O11 Dynamic EVO XL (White)
be quiet! Dark Power Pro 13 (1600W)
Thermalright TL-C12C (120mm)
Thermalright TL-C14C (140mm)

All Comments (21)
  • @shodan6401
    Just by activating Synch All Cores you have fixed the problem. What cooks these CPUs is when a single core, even for a background task, asks to boost to 5.8Ghz. Then, to reach that freq. it is given 1.55, or even 1.6v, depending on the individual CPU. THAT'S what's killing it. Even though the chip has power limits, when only one core wants to boost, all of the power is offered. This is a feature just so Intel can get a good CineBench score by reviewers. It's totally stupid. Lock those cores together and you're good.
  • @enilenis
    All 13th and 14th gen CPU's are duds. Intel probably realized that they can't make silicon any faster, so now they want CPU's to be replaced like light bulbs. Every few years they fry and then you're expected to buy them over and over to just keep going. Technically, they should recall everything they've made over the last 2 years, but no chipmaker can afford that. Only reason they're not bankrupt yet, is because not everyone follows tech news. And as chip begin to fry in increasing numbers, class action lawsuits will fly. Since Intel can't offer non-defective replacements (all of them are equally defective), they have no options. And people built entire machines that they now can't use, if they switch to AMD. Additionally, before the CPU problem was admitted, a ton of RMA's have been filed for motherboards, power supplies and GPU's falsely blamed for blue screens. Turns out Intel caused people with good hardware to swap it out. They owe hardware retailers and manufacturers compensation for lost profits.
  • for people that are afraid to not degrade the chip i always advise them to do a manual vcore cpu overclock, step by step and see how their frequency steps scale with voltage until they find out where the cpu needs a lot more vcore for the next step in frequency. Thats the only way to see if you're falling off the efficiency curve of your silicon quality and also you know that holding back a bit from that you can't harm the cpu (at reasonable voltages and temperatures) For example. (prime95 small ffts stability test) you need 1.2volt for 5.5ghz all core Then for 5.6 all core 1.250v for 5.7 1.3v for 5.8 1.35v for 5.9 1.42v for 6ghz 1.52v Your efficiency curve starts falling off at 5.8ghz and if you want to not degrade the chip you have to either run the cpu at easier workloads or and a lot lower temperatures. or never go over the efficiency curve.* I can't imagine a silicon to not degrade at 1.5+ over 50c maybe a rly old 45+nm silicon from 2000s (or fx amd😂)
  • @decibal7532
    No issues here with my SP106 13900K. Just undervolt and lower clock speed. Do not use the motherboard built in AiOC, that is what is ruining CPU's. I'm running 5.5GHz at 1.34V and an offset of .05V. AiOC will run your CPU @ 1.45V to 1.5V+ which is crazy.
  • @cademckenna863
    If you don't have a LGA bracket i'd suggest that, got my temps down about 10 degrees under full stress
  • @xsparky0168
    altho the wild support is nice, the nolife copium comments is kinda...... depressing ngl 😅
  • @dramion5486
    I have a i9 13900KF that I got right when it came out. The only thing I have done to it was set a offset of -0.060. I have never had one problem. No crashing, vcore never above 1.38. It's baffling to me, almost as if the more recent batches have an issue.
  • @shodan6401
    THE *FIX* PLUS EXPLANATION OF THE INTEL CPU PROBLEM: This is not a mystery or drama anymore. From J2C's BSODs, Actually Hardcore Overclocking, + many more, we've now observed multiple examples of EXACTLY what the issue is: The crashing and damage is caused when a single core goes into boost behavior. It can be two cores also, but it's the same problem. The CPU has a max voltage. But if you make ALL of that voltage available to one or two cores that want to boost to 5.8GHz, then they will get 1.55 or 1.6 volts. Whatever voltage is required to reach that frequency. That damages your CPU silicon. And that is an intentional design choice made by both Intel and AMD. It's not a manufacturing problem. Why doesn't this issue affect Xeon chips? Because most Xeon CPUs don't have single core boost clocks. When all cores boost together, the max voltage enforced by the chip's microcode is distributed across all cores, and no single core can get crazy high voltage. Anything over 1.5v has the potential to damage that core. Or possibly the cache package strapped to that core. The "FIX" is to lock/synch all the cores together, eliminate single core boost. This stops a single core from getting 1.6v. It's simple. If you already have damage, you may have to reduce your all core max boost frequency multiplier, because the damaged cores can no longer reach those freq. and will crash your PC. First, lock the cores together, no single core boost. I would recommend also putting a cap on the max voltage. Maybe 1.4v max, but because there's overshoot, personally I'd go lower. Not to mention the challenge of temps if all core turbo is too high. Validate what works with your CPU, what is a reasonable match between freq., voltage and temp. That's it. You're done. And no more damage. No more shouting into the void. The point of single core boost is stupid anyway. It's literally to get on the top of some reviewer's chart. But since both Intel and AMD allow their chips to do it, and we're at the limits of what the substrate can provide, we will keep seeing this. Lock/Synch your cores together, so when Win Update suddenly runs, a single stupid core isn't provided 1.6v so it can reach 6GHz. Problem solved.
  • @winstinwin7650
    Bro you got an SP rating of 110! I’m over here with 96 sp, bloody Intel was using my chip as a door stop before it got to me lmao FML
  • @BrockSuire75
    If your system is already crashing with Intel 13th/14th Gen you need it replaced. Im a hardware engineer and see many red flags with this behavior on these chips. Intel needs to replace all of them because large server companies are losing money & switching to AMD Server chips quickly. If I had one of the chips I would RMA it ASAP. Only bad things can happen to Intel from this point forward without fixing the problem. Thanks
  • 6:21 with 110sp chip it should use best case scenario for svid behaviour. Unless the table is stupidly low or your motherboard has rly bad voltage regulation.
  • this is what i did to my 13700k immediately when i bought my 13700k: disable hyperthreadding, set max core voltages to 1.35v lock down clocks to 5.5ghzP 4.4ghzE. thing runs like a charm and wont be peaking 1.4xx+v during idle.
  • @Toopy_GG
    any dx12 game would instantly crash for me but the recent bios update for my asus z790 e gaming wifi has seemed to fix the crashing issues. has run flawlessly for around a month now and i no longer have to force dx11 or undervolt the shit out of my 13900k
  • @AvitaBeckling
    you add 3 more fans for a push/pull on that 420 artic AIO get about 9-12C lower temps recommended if not.
  • in tweakers paradise that you said NOPE .you can actually go and set max vcore limits so your motherboard doesn't auto vcore shoot to the sky for not reason. your settings are rly bad optimised for performance, like a LOT . your llc at 1 have massive vdroop and terrible transient response.( 3to 5 depending the workload has the best transient response) also the settings for vrm efficiency has nothing to do with how much voltage the cpu receives but for how clean , fast and how stable will come. Your settings are for your motherboard's lifespan that actually never have issues if you got a decent motherboard. mosfets have temperature min limit to 115 if you check it😂 you can't set it lower!
  • @Waldherz
    You can just HOLD the del key.... USB keeps sending check requests. No need to spam.
  • @ultra_sound.
    My opinion: dont lock your cores. You have a really good cpu, sp 110 for 14900k is the best you can get. What is killing these cpu is the ring, not handling the voltage for p cores, e cores, and the memory controller. By your sp score, your vid table has likely very low vids for the p core frequency. I would let the bios on intel default, remove-enable limits and work on LLC, Dc and Ac LL. You will have low VID and will never go above 1.4v if not 1.385v based on your SP. You have a beautiful chip to lock at 5.0ghz.
  • @cademckenna863
    my I7-13700KF be pulling 300Watts under full load. too damn much wattage
  • @mikenone4077
    Not sure why im here but i have been an intel user for years and just decently switched to my first amd just because i wanted to try something new and different and what do you know a few months after i build my pc i hear about all these intel problems. Wow i got some really good luck somehow lol.