As a 3D artist, I'm flabbergasted that any performance mobo vendor would say Cinebench is unrealistic. Cinebench is literally recreating a typical workload I have, but for shorter durations. I'd like to think vendors selling me performance hardware to do that job actually design to be stable with that load.

The fact that mobo manufactures are complaining about it being cinebench shows how little they care. Like Jay said, it's a thing you can run on your system, and the system shouldn't blow itself up over it. If it blows up with no user changes, that's on them.

I had exactly this problem with my i9 13900k with games, especially UE5 ones, crashing way more often than they should. Also, my Bambu 3d printer slicing software crashed consistently for larger models unless I set CPU core affinity. Finally, while trying to figure out why Helldivers was crashing frequently on me, I stumbled across someone on Reddit mentioning this issue, and the same fix mentioned in this video. After applying it a couple of weeks back, everything has been running flawlessly. Not a single crash of any software since.

thanks for the video. havent had problems yet but went thru and enabled the setting.

As you and others have pointed out, THIS SHOULD BE THE DEFAULT. The most stable limits recommended by Intel *SHOULD BE THE DEFAULT SETTINGS OF THE MOTHERBOARD*

How is Cinebench *not realistic*? Its literally the render engine of Cinema 4D and 3d render is #1 use case for a consumer-grade 24-core CPU.

Thanks Jay. Brand new i9 14900, water cooler, on an MSI board, kept maxing out temperatures on minimal loading. I checked the cooler, I changed the thermal paste, then started looking for answers. Thanks for this and a few other videos. It wasn't me, it was the mother board settings. I've been tweeking, but I'll go in and check the setting you posted on this video. I agree, stock, out of the box settings should have settings at Intel defaults, not overclocked settings. Thanks again.

i tried your tips and it helps quite a bit. the issue stopped coming up when playing RoboCop but i also did open a support ticket with intel to see if they would do a CPU replacement in case the CPU is failing. thanks for you advice!

My Asus board had a bios update that placed the Intel stock settings as an option. You know, the thing that should've been there from day one.

Silly Jay, Airbus doesn’t have a button to eject passengers!! That’s Boeing….

As a geek squad technician, lots and lots of clients have come in with this same issue, I've been doing the same thing for 13-14th cen CPUs, Drop the core volts to 1.4-1.3 volts, and dropping the p cores to 5.7-5.4 ghz. Most of not all clients haven't had issues since. If you want that 6ghz the CPU advertised. It's only going to be on 2 p cores or less! Or you'll need a 480mm liquid cooler or dual radiator custom cooling setup to handle 6ghz most cores. E cores drop those to 4.4-4.3ghz These settings still get good numbers achieving 38k-40k points on cinabench r23. Hope this helps and good luck.

I'm going to test this method, I sometimes have completely random blue screen problems (usually when starting or closing a game). Thanks for the information and explanation!

THIS is exactly the issue me and my friend have been getting. He had this issue for over a year and just recently getting to fix things around. Overwatch kept crashing because it was one of the few games that pushed my CPU really hard (over 300fps) and thus causing crashes. I set my max W to 230 and removed the MSI default OC crap and been working wonders ever since. Awesome to see this put more to light.

apparently its not fine to run your cpu at 100c+ without power limits. I respect amd on thermal limits. 85max take it or leave it

For my 14900k and my Asus ROG Strix Z790, enabling the intel default power limits didnt fix it and I had to go in and manually set the Long Duration Power Limit to 253 and short duration Power limit to 253 and that fixed it. I was getting the Video Memory Allocation error on Remnant 2 From the Ashes before and updating those settings fixed it. Regardless, thank you so much Jay, I couldnt have figured it out without your video because I was scared to change the power settings and your explanation helped me figure out which ones to change

Thank you for this video. I just got myself a i9 14900K, and I had that error a few times. And I had no idea of these settings on the mobo.

less than a 5% performance dip but a 15C reduction in temperatures plus more stability... i'm all for it!

The situation reminds me of the Coppermine Pentium III 1.13 GHz for much the same reason: Intel was losing ground to AMD back then, too, and pushed its silicon too hard to keep up. Electromigration was a hot topic at that time because of the high voltage being used to push the silicon.

same problem here thanks for the video!

What a great video. Thanks mr 2 cents. My PC is now not running fans like a Dyson to hold the cpu coolant in mid 40s and is now super quiet and upper 30’s all simple by taking the extreme factory overclock off. Keep up the good work

Board manufacturers are also misleading about validated XMP speeds. You need a unicorn CPU to run 8000 memory stable without any additional tweaking.

Was thinking of upgrading to 13700k but got convinced to go 7800X3D instead. Hopefully it’s the right choice

Thanks for this. Although I ended up going back to the mobo overclock, this was very useful info which I didn't realise I needed to know. With some experimentation, the simple background info provided by Jay helped to clarify what I had previously Googled. Very, very happy with the results.

It's worse than you think, Jay. These over the top power limits seem to apply to non-K SKUs as well.

Great video, thank you for these informative videos! FYI I have a Z790P wifi motherboard running a 13700 cpu they just released a beta bios version that loads Intels factory settings. I have updated the bios and ran cinebench. My temps were low, no stuttering in games after the update. Hoping motherboards keep listening to the end user to help conserve our CPUS.

This is a problem that has occurred on two machines that I and my colleagues have built for our customers with a 14th gen i9 CPUs. They kept crashing Windows during rendering AutoCAD (which is the primary thing they got those expensive machines for). On one of them it got to the point that we had to replace the CPU because it was so unstable from that stupid out of the box overclocking. Thank you for this video, it'll serve as a great guide on what to do on our next order.

Hey, thanks a lot for the video! Just to make sure, should I update my BIOS to the latest version and then turn off the "Enforce All Limits" option?

Wow man you really helped me I honestly appreciate it and it was easy I was nervous to mess around in the bio

good stuff Jay, i was having this problem after doing a bios update and setting the cooling to "water cooling"(which i have just in AIO form) what i didn't know was this was uncapping my voltages and making my CPU hit 100c anytime any load was placed on it. I'm pretty decent with computers and overclocking and it took me about a month to find out that this is what was causing my temp issues. The average user will probably never even find this hidden setting, it's just unacceptable. Keep up the great work man! ❤‍🔥❤‍🔥

Let's destroy your pc it's worth 2%.... I'd rather slow my pc down to and be happy for a decade. I'm not spending 1000 bucks every 3 years if I don't have too.

thank you, Jay (and team), this is valuable information. I would have started looking at the video card with "video memory" error. As always, thanks for all you do 🤓

I had an i9 and asus motherboard blow up last year, I assumed it was a bad cooler. Thank you for making these videos and keeping us updated on these things.

I recently helped a friend walk through the issue causing crashes with hyper threading turned on. I9-14900k crashes reported as an Nvidia compute library file, but it cleared up after hyper threading was off. Weird, weird issue.

THANK YOU for reporting this. I'm chatting with Intel Support now and it's extremely frustrating! If they don't smarten up and do the right thing I'm done with their processors moving forward.

Had same issues you spoke about for my 14900k during first 3 weeks of 14th gen release. Wish i knew this information beforehand. z790 hero / z790 strix e Artic liquid freezer II 360 Used two different motherboards and two different 14900k chips. Both scenarios produced the same results. Cinebench and most games would crash upon starting and cpu high spikes. Everything was tested stock settings out the box. Returned both chips and went with a 13700k. Everything ran stable.

I have a 12900k and I did what you said to do for the 13th and 14th gens and got the same results. I lost points too but instead of always showing 99 degrees during the Cinebench test I now maintain around 67 to 72 degrees. Worth it. Thanks! :)

I would love to see benchmarks on the correct settings for intel. I wish all reviewers would have catched this a long long time ago.

For the 13700k I disabled the enhanced mode, set PL1 and PL2 to 253w, set adaptive vcore and I used -0.100 for the vcore offset. You can get away with the vcore between -0.080 to -0.050 for the vcore. I went -0.100 to have a cooler system. I tested it and got 31k on Cinebench R23 (same results as stock) but about -15 degrees less compared to stock. Very happy about it. Side note: I swapped the stock Intel CPU holder with a proper on from Thermaltake when building up my PC to avoid cpu bending and get better cooling.

Had instability issue with I9 14900k, used this and I can now pass a cinebench run with crashes. CPU would ramp up to 90°+ almost instantly. Now stable around 66°. Cooler is a Ryujin 360 AIO.

Incredibly well done and informative video; thanks.

I have a 14900k on a Gigabyte z790 master x and been having this issue for months since I upgraded to this setup and had no idea what was going on. I finally found info that you just explained in this vid this past week and yes I had to turn my 14900k down to 55x and now I get no issues loading/playing UE5 games or any crashes in any other games. Which like you said is crazy how we have to run the K series processors slower when we are supposed to be doing the opposite with them.

I had problems with an i9 13900K on an ASUS ROG Hero Z790 from day one. I updated BIOS, changed RAM, formatted the PC, and the BSoDs continued even with undervolt or with the configuration that you have shown in the video to leave the limits as Intel dictates. In the end I made an RMA to Intel and after giving them the requested information, they replaced the CPU covered by the warranty in less than 15 business days. So bad for the CPU headache, but good for the Intel replacement system.

Thanks Jay. Very good topic! And I'm very glad to see you well!

12:45 - Well changed Turbo Power Limits to Enabled. But it still shows PL1 Power Limit: 253watts and PL2 Power Limit: 4095 Watts. GIGABYTE B760 AORUS ELITE AX board.

Back when I was building servers for a living we tested all machines with a mix of benchmarks, and test programs that pressed the hardware to the limit. Most of that software came with a warning that it was not a realistic load for the computer but that was fine for us. It was real software doing just what was possible for the hardware. That it was programmed to press the hardware to the outmost was perhaps not realistic for normal programs but it was definitely reaching for the possible limit for the hardware. If the computer could survive this extreme load for 24 hours in a 40 degrees Celsius room without logging a single drive, memory or calculation error then it would survive any "realistic" load in more normal temperature. My point is test software is not intended to be "realistic", it's intended to be as nasty to the hardware as possible. The hardware should be built to safely process whatever software it getts to run. It's not enough that it works under optimal conditions or in cool environment. It has to be better than that!

With an Intel (MSI) the one to look for in BIOS is 'CPU cooler tuning' it should be 'boxed'

Switching to Auto from XMP 1 and 2 (both were very unstable) and following this video's instructions, literally healed my worries. I even turned off fast startup and switched to balanced power savings. Thank You! I've been struggling to boot the pc after sleep and shutdown. I used to hard shut off just to reach BIOS. Thanks :) 🔥

Hey Jay, what chair do you use to game/work on?

There are people who compare 13(4)900K with the FX-9590, that's not actually a fair comparison. FX-9590 was putting out 220W into one of the largest die size ever seen (it basically takes up the whole IHS) using 32nm. While Intel is putting out 300-400W into a tiny 10nm die, the power density is so much worse. Even after 13 years you can still find FX chips that has run since than without degradation.

OMG THANK YOU When I get home I will update my bios settings..... I spent like $700 on a freaking cpu just to get non-stop crashes and "out of video memory" messages! I bought a new cheap replacement cpu that works for now, but I want to use my nice cpu i thought was bad! i think this will fix my issue. I tried so many things, cant believe this is my issue.

Thanks Jay! I have one question. I have a I7-14700KF and manually set my powerlimits to 253 watt on both. I did that already a few months ago as i saw the same crazy high limit. That it never reached. Unlimited was aboout 280 watts max. Still very high if you ask me. And i undervolted it by 0.06500V to have lower temps. I did not see about the amps and will check it. I also went to the latest bios version and have an Asus motherboard. What voltage is still safe or should you target not to go over with my CPU? Besides this i still am happy with my intel/nvidia combo but it is not nice to see all the issues and i don't want to become a patient as well. *After looking into the bios and setting it to enforce limits the amps went down from a crazy 511A!!!! to 280A by limit the Intel one! Now even with 253 watts powerlimit my 14700KF does not reach it by far. My performance went down by about 2.5k r23 all core to a 32.5k instead of 35k but the temps dropped massive! Went from 100c without any form of undervolt or default bios to first 90c with 0.065v undervolt now to 70c with about 180 watts power usage at 32.5k all core. No undervolt gives me the same 32.5k with about 10-15 watts more usage. So now i am confused a bit. Does this mean that it does not reach the 253 short tdp at all? Only by ramping up the voltage a lot? Or the amps in this case? Crazy if you ask me. The difference is day and night.

Just got a i9 14900k.. glad I found this! Thank you.

I remember on old bios we have “load default settings” and “load optimized settings”

You say that the MSI equivalent is EZ Mode -> Game Boost but mine was off by default. What actually affects the set limits (as far as I can tell) is your selected cooler, in the bios. When you first start up, it will ask if you have a box cooler, tower cooler, or water cooler. If you select water cooler (because most nowadays are running an AIO) it will set it to completely unlocked limits for both short and long (P1 and P2) along with something like 512A. Setting it to box cooler will set intel limits 253w p1/p2. Tower cooler kind of sits in the middle at 288w p1/p2. I find though that it still wants to exceed those limits with the tower cooler setting, just like the water cooler setting. So, set your cooler to box cooler, and you will be enabling intel defaults. You can also adjust the Lite Load Mode from auto (which on my z790 carbon with an i9 14900k was Mode 10 by default) to something lower like Mode 6 or 8 to drop the voltages a bit more, but this shouldnt be necessary for intel spec because you should realistically be running in the 60-70c range under full load with adequate cooling. Additionally you can adjust your Load Line Calibration to change the Voltage droop when the cpu goes under load but thats kinda going into the weeds. Just set your cooler to Box Cooler for intel limits. I've never seen Game Boost/Creator Genie enabled and honestly, I have never even enabled it in the past to try it out.

Yeah. By doing that "little" change and of course updating BIOS previously, I could fix the problem after all these months. Thank you so much for the help!

I’d be interested to know where one can find all the power delivery information for all their processors like the one in the image from Falcon Northwest

Man, I'm planning an i9 build right now, this is invaluable info

I had this issue since the launch of 13900k. I was going crazy till years later everyone is having the same problem as I did.

@jayztwocents where would i find said profiles for the i7 13700f? please, and thank you

Ok JayzTwoCents, but I have a question: what about Core i9 laptops, like these ones from XMG/Clevo with Core i9 CPUs? I'm already well aware that the first thing I need to do is undervolt it using Intel XTU (since the BIOS on these machines are limited). What do you recommend on such cases? What will you do if you get one of these toasters? Thanks in advance

as a rule for my old work, any PC used as a workstation always had the auto settings turned off, stick to the limits and just use it

I recently had this happen on an Intel 7th generation i5 processor!

Very nice, thanks for the tips. I just started using an ASUS motherboard and can't figure out how to set limits on my core speeds. Have you or could you make a tutorial on how to limit my max power usage? Because AI Suite 3 always seems to want to set my power to the max and my CPU is cranking up to 70C regularly.

Happy done this tip on my 14900ks and the temps are great and the cinebench score is the same thanks

I cannot thank you enough for this video ive been ripping my hair out looking for an answer on why my pc kept crashing with the video memory error. I had reinstalled windows multiple times, i was blaming armoury crate for being a load of garbage. I got so frustrated I thought I had a bad connection and took my entire pc appart and redid all my connections reapplying thermal paste to my cooler. I am so relieved to have found the answer finally thank you so much Jay.

Intel needed the benchmark wins more than board vendors, so they for sure liked those power limits maxed out. Or else they would restrict them like they did in previous CPU gens.

I learned about this issue a couple of weeks/months ago and immediately noticed improvement once I set those limits in the BIOS. Way less stutters in games, CPU running cooler... It's really bad that they do this but glad it's finally getting some light because it solved a lot of issues for me.

This was the video I've be looking for, thank you!

The wild part about this is I was getting that specific crash on my 12700k rig. Never could figure out what was wrong. I recently built a new 14700k rig and it does not do this at all which is wild giving this new found knowledge.

Now feeling happy with 7700x , early memory issues were annoying tho.

Oh jay ❤ always maximum information inside a small amount of time thank you very much 💖

Would it also not help to go in and tune it for undervolting to help with the stability issue? or are core clocks the only thing you should be concerned about when battling this issue?

Hi, 14900K user here. So the story goes: First 2-3 days stock MB settings with 360mm AIO the CPU was stable. Then, started messing with XTU AI and optimizer all stable and R23 and Cinebench 2024 high scores for one day. The next day booted the system and started crashing, restored the MB stock settings, and everything was fine. A day later crashes again, engaged the CPU limits (Intel POR - Gigabyte Z690) everything was fine. A day later random application crashes, random shutdowns and blue screens with Intel power limits. Counter-Strike 2 is crashing without FPS limit, but as soon as I set a 144fps limit the game runs fine. Battlefield 2042 is not playable at all (this is with Intel limits, I haven't tested it with MB defaults). I'm going to RMA the chip and I hope that I'll get a good chip this time. Good content Jay!

Jay needs to learn the difference between a Perimeter and PARAMETERS

Bro thank you for this I genuinely was getting confused I'll be adjusting these settings when I get home since I am getting these errors on my rig I thought I had a bad 4070ti super

I'm really glad you made a video on this topic. I'm prepping for a new system in the coming months when Microcenter opens up in Charlotte and I've been on the fence with the 14900k since hearing about the crash issue. Coming from a 9900k, I'm wanting to stick to the i9 chip and have all the extra juice I can but I'm questioning the choice now... ugh, hard decisions.

Really happy with my AM5, pcie extra lanes, lower wattage, first time I went with team red

Well, well, well, how the turntables.

Will be putting together an i9 13900k build this week, will do this first thing. Appreciate the info and heads up on an avoidable issue!

Thank you so much for this video it fixed my issue you're a lifesaver

My 13900ks broke. My 14900k is braking now. I spent 5k to have a great computers for years to come. All I've had is trouble and tweaks.

Much like the Ryzen burning issue AMD controls what their partners can and can't do with their products. Intel controls what their partners can and can't do with their products, Intel is to blame here just like AMD was to blame here, Intel enjoyed MCE giving them better out of box results compared to competing products because it wasn't a 'stock-stock' comparison. Now it's biting them in the ass.

Same thing with mobo manufacturers for am5 there were BIOS issues where the core cpu voltage was too high or allowed vendors to increase it. This resulted in some 7000 series CPUs (the X3D models mostly) to overheat and buckle/crack the substrate on the CPU. That is why I am still using am4 don't know if they have fixed that for Ryzen 7000 series now though.

Amazing video thanks

Good timing award on this Jaz because this exact issue has been occurring in just the last few days for me. I thought something was strange about it because I have a 4090 and 64 GB of DDR5.

Timing is everything, seems last year when AMD was having the motherboard 7 series AMD issues, all content creators were saying in their own builds they are doing Intel cause reliability. Almost around the same time a year later, its flipped. Everyone has issues. It happens.

Hello, I saw your video and checked my settings, I have the 13900k, and were set as the numbers you shown. Asus has Bois update ver 3501 that corrected to the numbers shown to be used, I updated and now their correct and have changed back to let bois optimize. Running well and a little cooler. Thanks

Got my 13th gen, i7 earlier this year. Dodged a bullet… Thanks for the great vid, as always!

14900K owner here, no crashes/errors yet. When people said this chip was hot, I didn’t think it would be too hot. Out of the box in Cinebench temps instantly hit 100°. I’ve got two thiccc 360 rads with an Eisblock on the CPU. My system was also too loud during gaming (and I have all Noctua). I had to drop the voltage to 1.15 and max power at 225 (I don’t need 6.2Ghz). Sure, I got a lower score in Cinebench but in gaming I honestly can’t tell the difference. I also find it funny that I can call my 3080Ti a cool (temperature) product.

I hate to throw a monkey wrench into this discussion, but the "out of memory" error is NOT just happening to the 13th and 14th gen CPUs from Intel. Systems as old as my 4770k can experience this, and it appears to be at least partially game specific. A really good example of this that I noted over on Steve's video regarding instability for the 13th/14th gen Intel CPUs is with Diablo 4. A modern game (with a terribadly optimized engine in many respects) that is having that very same "out of memory" error. The thing is, the Fenris logs for that game clearly show that the game didn't hit a *system* RAM wall, it hit a *virtual memory* wall. That is, for some reason, Diablo 4 as of the last few patches since roughly the beginning of March, are not swapping textures to system RAM like you'd expect, but merely dumping them into the VM pool. On my 4770k system with both a 1080 Ti and my 4090 in as a test (the 4090 is in there because I needed to test the card before its return period was up and hadn't yet put together my more recent build for it due to a bad back and having to move a Cosmos II case), the same out of memory error occurred. Every single time the result was the same: the game noted it had run out of *virtual memory*. When the game crashes like this I typically have 10-13 GB free of the 32 GB RAM I have in this system. The 4090 made a very interesting test scenario because of its absolutely massive amount of VRAM. In the case of Diablo 4 it appears that the developers kind of did a massive pendulum swing from previous patches. Prior to 3/5/2024, D4 was doing garbage collection at such an aggressive rate that it would induce frame pacing and animation time errors (stuttering/hitching). So as of at least the 1.3.3 patch it appears that frame pacing is smoother, but apparently at the expense of not swapping textures into RAM, but for some reason VM instead. Windows' default VM is going to be nowhere the size of your GPU's total VRAM in 99% of cases. The only solution found here was to *increase* the Windows VM size to ≥16 GB, which is an absurdly large VM size, as that then locks off that much of your boot SSD, which usually tends to be the smaller of any drives you have installed. Regardless though, users should never have to adjust the VM size because a game isn't coded properly. Yet here we are. So you might want to have the folks over at Falcon Northwest give that a try and see what the results are. It's possible we're dealing with multiple different out of memory issues across various games currently, but the timing of this is really suspect relative to when the D4 1.3.3 patch came out and it kind of has me wondering if developers aren't being super lazy and/or sloppy with how they manage memory and/or do garbage collection in their games.

Great video Jay, I will probably be making these changes on my custom builds to protect users from 'themselves'. And another observation is why is Intel not taking legal action against the motherboard manufacturers? I wanna put NOS on my V8 sports car and when I blow up the motor I expect warranty coverage (okay, maybe not the best example...but you get my point)

With MSI, changing the setting in EZ Mode didn’t change the numbers after restart. Had to change them manually. Thanks for advice

I'm happy going to 7800x3D this summer...

I got unlucku with a 13900k it shows out of video mmory i gave it back to the pc builder company where i bought from now they will change it to 14900kf since intel do not make 13900k anymore. I hope i get a non defective cpu this time.

Had similar issue on i7. I disabled Asus AI OC and the problem seems resolved. I am seeing rare blue screens, with unknown causes with default setting. I will freely admit, I thought Optimized Defaults were safe values. This was a good video, thanks!

Thanks for the info :)

It's like we heard this same exact issue on the 9,000/10,000 series intel chips? Time is a flat circle.

So Cinebench results are tainted. CPUs are being run out of spec.

Didn't think I had stability issues but I guess these qualify. My cinebench would crash and not complete the 10 minute run. My RAM wouldn't do the 3600 like advertised and had to do 2933 to run cyberpunk and God of War or the games would crash. After I changed the settings to what you recommended I put the RAM back to 3600, because I was curious, and Voila!, the games ran fine with 3600 MT/s and my cinebench went through the full 10 minute run. Cinebench score went from 40144 to 39506 after I changed the settings and it ran at 253 watts with about 5300Mhz clocks sometimes 5400Mhz Equipment used: Intel 14900k Gigabyte Z690 Aoris Elite DDR4 RipJaws V 3600 MT/s Hope this helps the other guys. thanks for the Vid Jay!

"you wouldn't have your car like that" Well, my mate had a v6 ford mondeo back in the day, where the fix to it pinking was indeed to change the timing such that, it had "less hp". Still wrong of course. I'm so glad I stuck with a 13500 for my recent 'super upgrade' ( from a 6700). A friend got a 13900 for his office pc (which he is running lots and lots of tests on, so uses those clock cycles) and he's had all sorts of problems with it. the info and fixes from videos like this has been really handy.