Intel's CPU Crashes Explained 

hey yall, for me it is "why not both" -- on gaming boards The out of box defaults really are insane. linus' take overall is reasonable. and I looked over the script beforehand. the uncertainty on gaming boards is a big deal and why I went digging on more "conservative" boards like in the datacenter. which are also imperfect! should intel have caught any microcode issues sooner? absolutely yes. have motherboard makers pushed voltages directly or indirectly to ruination? also yes. Is it up to intel to make the situation right? yes I am hoping it wont take a lot of pressure and media attention to get Intel to do right by customers... how we got here is more nuanced than just saying intel pushed too far too hard..and its super important to see the whole picture of the situation ps update your bios!! and update it again mid august! even if you sont currently have issues. :)

The problem if your chip is damaged, no amount of bios fixes or anything short of a new chip will fix it.

You guys are playing softball here - Intel is absolutely responsible and you don't need to be nice about it. - They shipped a defective product with the voltage regulation being completely buggy. This would have been caught with the extensive testing Gamers Nexus, JayzTwoCents, Hardware Unboxed, several PC building companies, etc. performed. - Intel was clearly more interested in "number go up" in their implementations than figuring out if it was actually feasible and sustainable. - The oxidation issue is ridiculous - again, should have been caught and quality controlled long before it ever hit the market. - Not providing any real guidance to motherboard manufacturers - leaving them to their own nonsense of overvolting and blowing up systems. Let's not play games - these processors are hundreds of dollars and with all this news Intel completely went radio silent until their bottom line was under pressure. The C-Suite pushed out broken processors and expected the consumer to just deal with it. They might be fixing it now, but they knew about this for almost a year and are only now acting like this is a shocker. I see this as consumer abuse and neglect, and no one should stand for it.

LSS - Linus Slim Shady

6:43 'Prevent damage' ? You will never know what damage has already been done once you used the CPU before bios update. The statement should have been 'to prevent possibly further damaging your CPU' This statement makes consumers think their CPUs are ok once the bios update is undertaken when the damage could have already occured. You guys need to encorage RMA and not just this band aid solution.

I'm sorry this is 90% on Intel, poor microcodes, oxidation and poor definitions of "default" on the information released, so MB manufacturers still have room to push "default" settings.

No recall, replacements, returns? Gonna be a class action over this.

With 15,000 fewer employees, I don't think that oversight of the motherboard is going to improve anytime soon.

I fixed the problem 4 months ago, I bought an AMD chip 😊

Oddly defensive of intel. Fact is, Intel failed to produce clear guidance on how to run their chips safely. And considering the microcode issues, it seems that they didn't even understand it themselves. Sure the motherboard vendors also verged on recklessness, but that situation occured and then wasn't adequately fixed because of Intel's failure to make clear statements about safety criteria. And now they are using that lack of guidance as a smokescreen to distract from their responsibilities.

Why is Linus placing the majority of the blame on the Motherboard manufacturers, when he says in the video that devs are replacing their Intel servers with AMD, and unlike retail MB's server motherboards follow much stricter guidelines and not overclock the CPU's. something Wendell went into quite extensively. I think the reality is very likely Intel knew this was a issue (though maybe not the extent of it) but because they also knew AMD's new chipsets would beat them in gaming performance, they allowed MB manufacturers to push the CPU's beyond acceptable limits, hoping the situation would not be as big of a issue as it turned out to be.

Can the Real Slim Linus, fix my pc, fix my pc, fix my pc?

We have around 50 production servers running on 13900 chips. Every other day we had unpredictable errors on most of them. Right now we are migrating to AMD-systems.

You literally cannot fix it. Intel has stated that if you are already experiencing crashes, there is nothing that can be done. Your chip is done.

I'm glad somebody else said it too. This is NOT motherboard problem, this is a 100% an Intel problem. Sure, Intel have been blaming motherboard manufacturers as part of their "damage control", and the BIOSes have been playing a bit loose with their settings. However this is because Intel's specifications aren't exactly clear, so motherboard vendors in my opinion have been doing the best that they can. Intel, you should be ashamed and I'm just glad that I'm using an AMD CPU and don't have to deal with this shit. I also feel sorry for those that are going to suffer, most not even knowing if they have the problem or when it's going to rear it's ugly head on their system.

This sounds to me more like an Intel Problem than a Motherboard Builders Problem.

This video seems weirdly focused on the mistakes of the motherboard manufacturers. Yes, it's true, especially Asus and Gigabyte had some fucked up load lines, but that's not the main issue causing the instability. The motherboards aren't delivering such high voltages because of some error, but because the CPUs are *requesting* them via their VID commands. Some other effects like overshoot and measurement inaccuracy are then pushing voltages to even higher peak values than the requested value, but those are always a thing. Intel must have known of those and should have accounted for them, but they didn't. Because the chips apparently needed to hit 6 GHz boost for marketing reasons or whatever, no matter how much voltage you need to ram into them to hit that target. In short, Intel flew too close to the sun and got burnt. Blaming the motherboard manufacturers for Intel's mistakes isn't fair. Even the mistakes they did make originated from Intel's poor specifications allowing for stock load lines that were unstable which should be an impossibility.

Glad I went with AMD when they released Ryzen instead of the shovelware Intel kept pushing out. Just because they are throwing more watts at it to increase performances, doesn't make it better.

Intel is soley to blame here. They long accepted motherboards pushing the limits by not giving strict guidelines and not validating what their partners implement to get that last bit of performance to beat AMD while keeping plausible deniability. But Intel said everything is within spec as long as the overclocking bit is not set… you can’t fault board partners if Intel says it’s within spec. Also 13th/14th gen CPUs fail in W680 boards which run conservative Intel baseline values, Supermicro does not tweak out the last bit of performance and yet CPUs still degraded and became unstable. Intel rushed out Raptor Lake within a year because they needed something to bridge over until Arrow Lake is ready, and they pushed the ring bus hard to get that last bit of performance. Which is where the CPUs now seem to break. That pushing 350W through a desktop chip can’t possibly be healthy was clear for every buyer with half decent knowledge.

Kinda buried the lede there, didn't you? You spent the first 7:30 talking about how it's all related to power profiles on motherboards (what Intel has been blaming it on for over 2 years), and then at the very end say "Oh yeah, also the chips are defective."

Sounds like LTT next video will contain 'segue to our sponsor Intel' 🙂

Ngl, intel sounds like the boeing of the cpu market

I work at Currys in the UK in the repair centre here. We've had so many 13th and 14th gen CPUs come in for these issues, and we can't do anything to fix them as Intel blames motherboard manufactures and they blame Intel. It's really shitty for us, and beyond that annoying given us as repair technicans know this is an Intel problem entirely. If people have anymore questions, more than willing to answer

Why should the customer try to fix the crashing CPU? If the CPU is unable to hold its Stock spec Intel has to provide a replacement.

100% intel's fault and not the motherboard manufacturer

oooofff. wow, Linus. This is a truly awful and incorrect take. This is absolutely an Intel problem and failure. Sure the motherboard issues may have or certainly did make them worse but this is still on Intel. You giving Intel a pass on this is just stupid and wrong.

Intel is the only one to blame here they are throwing the motherboard manufactures under the bus for doing what they gave them the green light to do only after it caused problems.

My CPU (i9 14900K) just died on me 2 days ago. There's dark spots, that I take as burn traces on it. It had less than 6 months of use.

You missed out a very important point: In most cases, the part of the CPU that degrades is the ringbus interconnect, rather than the CPU cores themselves. This seems to be caused by a design flaw in Raptor Lake CPUs, in which the ring bus is physically not robust enough to handle the rate of data transfer required for high core-count CPUs, rather than by load line or temperature limits. CPU core voltage does also affect the ringbus, and higher temperatures increase the rate of almost all types of processor degradation, so this doesn't mean that the factors mentioned in the video don't matter at all, but ringbus problems seem to be at least as significant and should not have been left out of the video. It therefore may be more useful to disable the CPU's efficiency cores in order to reduce the stress on the ringbus, or reduce the ring ratio, rather than reducing CPU core clock frequency. This is especially true for gaming, which typically doesn't benefit much from efficiency cores.

Why are you pushing blame onto board partners for Intel's fuckup?

How did LTT take this as the mainboard manufacturers fault… this is literally the opposite of what everyone else is reporting. Likewise this could lead to intel taking the same road as NVIDIA and blocking messing with designs and overclocking

It is not possible to fix a physically degraded cpu.

This video is full of what were early guesses on what the problem is, a lot of it is not accurate and is still under investigation. I'm kinda surprised and disappointed LMG ran with this without a disclaimer.

Why does this video felt like a damage control in intel part? 😅

Not the first time Linus doesn’t know what is talking about xD. It’s not about load lines from mb manufacturing. It’s intel PL states and how they wanted to maintain higher clocks for 10 second to have better CB scores than AMD. Intel is providing everything to MB manufactures, there is literally zero fault on them.

Unlimited power profile was IN SPEC for Intel all this years since 8th gen ! To quote Intel from an interveiw with Anandtech in 2019, about the out of the box unlimited power profile: "Even with those values, you're not running out of spec, I want to make very clear - you’re running in spec, but you are getting higher turbo duration." Noting "vague" here ! Just now when the issues startd suddenly the motherbord makers are the evil ones who "step over the line" ??!!

LTT is more casual than technical these days. 1: 13900T and 14900T are also having issues on non overclocking motherboards "W680", these are rated to hit 65watts.. 2: The real reason for the crashes is more to do with the IO layer which is why you get data corruption on PCIe SSDs, and why lowering DDR5 speeds and disabling Ecores sometimes helps 3: This is a fundamental flaw in the IO layer and ringbus design, which is why its slightly older brother "12900KS" goes vroom vroom at similar voltages for years without fail. So in Short, Intel made space for more Ecores and voltages but in doing so also made the ringbus and io layer far more fragile. The only sure way to fix this, is to set a fixed speedstep frequency, disable all C states, and if some of the lanes are damaged move to a South bridge SSD to prevent corruption, secondly lower your DDR5 speeds, and set the voltage to a max of 1.4v if you're still seeing problems contact intel but in the meanwhile disable Ecores if possible. The less your chip has to micro switch and the fewer vdroops from c states or wake up happen the better it is for that fragile io ringbus. Good luck and good night from an actual Tech Lead, not some crypto scammer 🫡

So why did the workstation/server class motherboards used for game servers also get screwed up? They're specifically designed not to push the boat out too far for stability sake.

It's all Intel's fault - mostly - they admitted to it being microcode and manufacturing processes related - they're planning to put a patch out to fix it. We'll see how well that goes.

How much Intel paid Linus to blame MB manufacturers?

hard disagree with intel having "some of the blame". they are the ones with very loose guidelines, they are the ones that sold defected cpu with oxidation, and ultimately the ones that decided to react only when there was a media out cry. That is Mostly on Intel almost every step of the way. Gamer Nexus viewpoint on this matter aligns more with mines on this subject. i could see why this video got heavily disliked.

This video just is not accurate. The degradation of the Intel CPU's has been shown to be caused by the chips themselves requesting higher than 1.4 VID trying to anticipate Vdroop.

I was planning on my next PC build to be Intel. Yeah that's out the window now.

This is how you begin to lose integrity. I won't let this one situation sour LTT entirely in my view, but cmon, we all know whats happening here. I get why this company would want to beat around the bush with it, but we can all see the elephant trampling it. And thats that Intel takes 100% of the blame here.

You are shitting me. This is why my games have been crashing. Swapped my 13900ks for my old 12th gen 12900k. Dead ass stable in games now. Thought my gpu was dying.

I'll give y'all some credit... But you've missed another major issue... Intel knew about this issue long before it ever came out... And was denying B2B (business to business) along with B2C (business to customer) RMAs while they knew about the oxidation issue. That's beyond excusable. They've been caught, they know it, and now they're scrambling to slap a bandaid on the situation. I'm not saying AMD hasn't done any sketchy practices, as practically every business has... But this has me questioning if I could recommend Intel right now.

FACTUALLY INCORRECT on the part about motherboard vendors being at fault. When ANY processor manufacturer makes a product, they TELL the motherboard manufacturers the specs of the processor and how they want them to run.... So for example when AMD made the first ryzen series (1800x product example) they, via their microcode, told the motherboard how to auto-boost and do all the work. The motherboard vendors just ran the spec AMD gave them. THE SAME IS FOR INTEL. YES, motherboard vendors can have "overclock" modes that you can set which will auto-overclock products. THIS IS NOT WHAT IS CAUSING INTEL TO FAIL. You buy an intel 13th or 14th gen, run it completely stock, and it will fail on its own. INTEL ADMITTED TO HAVING TO UPDATE MICROCODE TO PREVENT THE ISSUE. And hardcore nerds have stated that this will cause the CPU's to lose about 10% performance. You are honestly going to tell me that EVERY motherboard manufacturer (intel's own boards, msi, asus, gigabyte, asrock, etc) all "worked together" to run these intel CPU's EXACTLY THE SAME in the EXACT same "out of spec" manor? GTFO. Truth is, INTEL was losing against AMD, and instead of having a slower product, they pushed it passed its limits in order to stay relevant. This bit them in the ass as chips are self destructing. It has NOTHING to do with motherboard vendors and 100% to do with intel being greedy and trying not to lose their ignorant customers that buy them "no matter what".... and when this microcode update rolls out fixing the cpu's and running them "as they should have been" in the first place.... they will all drop 10% performance.... And for the super ignorant that claim "intel rushed the product" no, they didn't "rush".... you honestly think they would scrap the entire cpu line because it didn't meet their speed goals just so they could get that 10% performance without destroying the chips? wake up and smell reality for once.... they didn't rush anything. they had a product and it couldn't compete vs amd.... they had three choices.... scrap it, redesign the chip, which would put them BEHIND AMD as 7000 series would be out, and by the time Intel had a new chip, AMD would be moving to an even faster chip.... option two being sell the chip as is, and lose against AMD (the most respectable one in my opinion) or option three, which is what they chose, was to overvolt the shit out of the chip to irk out 10% more performance so that they could stay relevant. they knew these chips would self destruct. THEY DIDN'T CARE. its not "rushing" when you don't give a shit about your customer. "our intel owners buy a new cpu every year anyway, so who cares if they self destruct, no one will notice" except people are more broke now than ever thanks to bidenomics.... so they are holding onto their cpu's for longer than they usually would. and they forgot that many intel owners are dedicated overclockers, who will run the cpu max speed 24/7 instead of simply using auto-boost. similar to how I ran my 1800x at 4.2ghz at 1.5v for 6 months until it degraded because i ran it over AMD's spec (amd said dont go over 1.45v and I went to 1.5v which was TSMC's "node" spec but AMD designed their chip in a way where 1.45 was max). so imagine how bad the degredation is for someone who legit OC's their intel product.... which is why people are getting fired. more than likely they simply didn't care about their job, and now they dont have one. good riddence. intel can grow and move on from this. its 100% their fault.

"13th/14th gen CPUs fail in W680 boards" . That sentence alone disregards this video.

Honestly I put very little on motherboard manufactures here. Yes motherboards should be responsible with how they adjust settings to compete, but when intel is giving absolutely terrible guidance on what is and is not a clear default setting I feel the blame should be 10% motherboards 90% intel.

Better watch Gamers Nexus video than this, Linus sold his soul to Intel

So glad I went with a 7800x3d on my latest upgrade but man this sucks for the consumer. Spending over 400 on a processor and then you have to buy a new mobo with it just for it to be faulty on arrival that ain't right

"The problem affects 13 and 14th gen chips" Me with 12th gen smiling.

I'm super frustrated. I built my new pc with a 13700k a year or so ago, and have noticed that my PC has been much more unstable than my old one (especially with certain games). I initially thought that it was a problem with my RTX 4070ti because graphics drivers, etc. have been blamed by almost every crash report that I've gotten even though I make sure to keep my drivers up to date. Recently it seems like it's been getting worse, with some files on my PC actually becoming corrupted. I don't want to have to be dealing with this crap. When I spend several hundred dollars on a CPU I shouldn't have to worry about there being issues with said CPU out of the box. Needless to say I have been thoroughly convinced by Intel that I should go with an AMD chip next time I upgrade.

This is way too soft on intel. So all the board partners happened to get it wrong simultaneously? The fact that it isn’t contained to specific motherboard partners tells you this is 100% intel’s fault. Watch the Gamer’s Nexus video for more info on how bad intel messed this one up

The oxydation issues cannot be fixed by a bios update.

The most recent Intel chip I bought was 11th gen. Based on this, the next chip I buy will be AMD. I don't even really care that they effed up. I'm more concerned about them shifting the blame and being so silent on the matter for so long. As a consumer, that is incredibly frustrating because there is nothing you can do about it other than waiting for a class-action lawsuit.

It's also happening on server motherboards, I highly doubt they are squishing performance like the mainstream motherboards do, so I wouldn't blame the motherboards vendors for this

Intel have been pushing their chips too hard the last few generations with their boost behaviour and the resulting voltages needed.

Is this out of date? Did you record this 2 weeks ago?

Making it sound like 75% of the fault is the MB manufacturers is rather misleading. If that was the case we would be seeing the same issue on AMD. Intel knew what it was doing by not giving adequate guidelines to manufaturers, thankfully it didn't end up being the get out of jail free card intel was hoping it would be.

I would point out... there is blame to be put on the MOBO manufacturers but they are not the issue... It is 100% the Intel Microcode... The instability is 100% to do with Intel and not their partners, at best the manufacturers are to blame for shortened lifespans on the CPU but the crashing and instability is all Intel...

Intel should recall all the cpu affected. Even if you have no issue you had a certain amount of degradation and the lifetime is compromised. This is ridiculous, usually in the chip industry you have to respect a certain mission profile that includes the lifetime expected in certain conditions.

So far, MLID was spot on, so I'm just waiting for confirmation that this has little to do with mainboard manufacturers at all, but is squarely just Intel's design is shoddy and cooking the ring bus.

is it Linus doing damage control for Intel?

Linus and Intel have a very close relationship. Not very surprising to see this moral hack shilling for intel and playing softball with them.

RIP to the guy who invested 700k into this mess of a company

step 1: understand you bought a defective product that the company likely knew about. step 2: proceed in the next steps however you deem fit.

I would LOVE to listen to the phone call happened between Linus and Pat Gelsinger (the CEO of Intel) right before this video :D I really don't buy it Linus, you say every motherboard manufacturer is uncapable to degrade an AMD CPU? All you say is invalid, why doesn't you apply the same logic to AMD or any other Intel CPU prior to 13th gen? Because those chips are faulty. And Intel should replace them. End of story.

this seems like a moderately dishonest announcement as an intel sponsored company, much better to watch steve for any real and honest information about the situation.

I feel like this video kind of misses the point. Also, Intel's vague guidelines could be because they needed the motherboard manufacturers to amp up the juice on their CPUs to beat AMD.

This all reeks of a cover-up. Intel didn't publish that little factoid about a silicon-level manufacturing defect until after blaming the motherboard manufacturers, and letting tech journalists run with the headline. The voltage issues exposed the defect, not caused it. And these microcode updates and performance throttles definitely appear to be a band-aid fix to buy time. If Intel can keep the flaw from causing issues until the warranty expires, then it's no longer their problem.

The problem is architectural. Not bios.

The root cause part of this video is just straight up false based on current information

Nice damage control, Intel, but it's your fault

GamersNexus did a full video, much longer, and more comprehensive, about this issue. He didn't soft-ball it either. Intel is to take blame in this, blaming the manufacturer for something they have been doing for so long. You mention it might be time to RMA it. GamersNexus has already revealed intel is not taking accountability.

Whenever I watch a video about this intel issue I can't help but wonder how many users out there have no idea.

This is a clear example of how you need to manage/navigate the relationship with a big sponsor of yours business. If you are harsh and lethal you will burn a bridge 100%. There's always a tradeoff with big things, remember.

I might be wrong, I just woke up a quarter of an hour ago from my evening nap, but to me it sounds a bit like damage control for Intel when you spend so much time talking about motherboard manufacturers and so little about Intel. Or maybe Intel isn't the one responsible after all, maybe it's people...

you know you messed up and are in damage control when wendell from level1techs is here trying to back you up.

Quite honestly it feels this script was wrote before Intel investigated the issue further and realised it was their CPUs and micro code at fault . Not what was initially blamed on the motherboard manufacturer. I feel the blame on intel was not fully understood and our knowledge of the issue has developed on from what appears to be focused on in this video. In my humble opinion

Long time viewer, but I can't even remotely agree with this take. Intel messed up, not anyone else, and the idea that we should be taking any of that responsibility from a company that is currently treating every effected consumer like garbage isn't okay. Intel did this, they messed up and it's them that need to fix it.

Steve is going to do a video about how wrong this video is. "Because another hit to their reputation like this is NOT gonna feel good." -Steve Burke enters the chat...

Linus and UserBenchmark sitting in a tree, s-h-i-l-l-i-n-g

Intel: Linus can you run a video to help us not look so bad? Linus: How much money you got? Intel: MY Maaaaan! Gamers Nexus:We’re releasing a video about Intel. Intel: No no no no no no no no no wait wait wait noooo

Wow, that was really heavy towards motherboard makers and pretty soft on Intel, along with actually being incorrect. At around 2:35 you claim it was the fault of some motherboards for not lowering the temperature till the CPU hit 100C. Yet on the Intel official spec pages, they list 100C as Max Operating Temperature for many of the CPU's that are now failing. So basically the motherboard was doing what Intel said it should, run the CPU as fast as possible (so that it can compete with AMD in benchmarks) and only slow it down if cooling isn't good enough to stop it going past 100C.

Its good to see you referenced Gamers Nexus. No hard feelings, that's the best way to improve, to take criticism, learn from it and improve.

Step 1 : Demand a Refund whilst you're withing Warranty Step 2 : Never Trust a RU-vidr to give you worthwhile information about any company they have affiliate profit sharing setup. Step 3 : Never be an early adopter.

i wonder how much dark cash floated through LTT this vid smacks of intel propaganda

I've had 5 BSOD's today within 10 minutes on a 14900K on a MSI Z690 board. Had to turn it off and go do something else. Experienced unexplained software crashing too over the last year. I've got press sample CPUs too so they're not retail.

The tone I got from this video is that MB vendors is responsible for 99% of the root problem that makes the chip degraded, and Intel has since 2023 begging the MB vendors to stay in line. So, Intel's only fault is that they're not strict enough to the MB vendors. Edit 2: So after a quick meal, I have to ask, why an issue this important isn't posted at the main channel (Linus Tech Tips)? The Linus Tech Tips having 16M subs will make the reach even wider, no? Edit: Mb to MB vendors.

Lmao, Linus still defending Intel with outdated claims after all the chaos unfolded

Just BIOS limiting the PL in watts doesn't tame the microcode's boosting voltages to get the single core to hit the out of box targets Intel was promising customers. They are the ones who went out of spec of their 10nm silicon. The microcode wasn't behaving incorrectly either, they made these boosting algorithms on purpose to eek out benchmark victories for their over-extended 13th and 14th gen products that they should have just not released until they had a viable product taped out on a better node. First they blamed their motherboard partners, now they have Linus shilling for people to nerf their CPUs before the problem (which they won't admit to) causes them more financial harm while they validate some new method of f*ckery to boost voltages in some different manner. They probably want to make the CPU touch 6Ghz for 1.5 seconds or something just so people can't sue them.

Linus trying to cover up what is clearly Intel's fault.

Damage control Linus lol. Pathetic.

There was a video about this titled "new intel CPUs are literally falling apart" by Low Level Learning, that determined it was a hardware issue, specifically a rusted wire in the CPU itself. Don't buy intel 13th or 14th gen CPUs and don't buy them used either - there is a reason for their used market presentation.

Intel itself admitted their VID microcode was faulty, that's not the board manufacturer's fault

my 14900kf started to have issues around a month ago. The first descendant wouldn't even launch. I would get crashes & BSOD playing league of legends and I'm crashing every other map in Path of Exile. Thankfully, IBuyPower is issuing an RMA for my processor. The pc is only 3 months old as well.

This video sounds more like Intel sponsored.

You are being way too nice to intel

Linus totally ignores the very long list of Intel's ongoing horrible anti-consumer behavior throughout this issue starting with the fact Intel is still selling these known to be defective products to unsuspecting consumers. The list of Intel's slimy behaviors is huge and well known in the tech space and Linus say's nothing? Broken CPUs might cost a small nation's entire economy to make right but, the massive distrust Intel, and some others, are sowing is a much bigger issue that has no fix. Likely its a fatal issue. R.I.P Intel.