News Intel issues statement about CPU crashes, blames motherboard makers — BIOSes disable thermal and power protection, causing issues

[Admin]

Intel issues a state regarding stability issues concerning 13th and 14th Generation CPUs on 600-and 700-series motherboards made by respective manufacturers.

Intel issues statement about CPU crashes, blames motherboard makers — BIOSes disable thermal and power protection, causing issues : Read more

 

[endocine]

So all of the settings that intel has been OK with, either tacitly or explicitly encouraging motherboard manufacturers to use. Why sell K SKUs and charge a premium for them if they are already at their maximum with intel baseline defaults, and why sell OC chipsets that allow this? Some of these are absurd:
"Using windows ultimate performance mode" <--is that a problem even with intel baseline defaults? Really?
"Disabling C-States" <--is that a problem even with intel baseline defaults? Really?
If those are a problem, then the CPUs can not be run at their baseline spec, its using too much power and generating too much heat, there are workloads that will max them out all the time.

PL1 and 2 have been set equal now, with unlimited tau, is that a problem now too?

 

[Amdlova]

Memory overclock it's another nightmare... my asrock motherboard says can overclock the ddr4 at 5000+ mhz Think what voltage at IMC and What gear mode... on my test from 3200mhz to 3600mhz incrase almost 10w from RAM and IMC voltage (blazing heat speeds)

 

[Alvar "Miles" Udell]

Perhaps Intel would have stricter guidelines for respective motherboard manufacturers to follow, given that such issues span for a few months before it could be narrowed down.

So their guidance should have been "When we say our recommended specs we really mean those are our recommended specs and not our "recommended specs", so you should actually follow them?" Do explain Mr. Shaikh how Intel should have been more strict in their guidance.

 

[TerryLaze]

So all of the settings that intel has been OK with, either tacitly or explicitly encouraging motherboard manufacturers to use. Why sell K SKUs and charge a premium for them if they are already at their maximum with intel baseline defaults, and why sell OC chipsets that allow this? Some of these are absurd:
"Using windows ultimate performance mode" <--is that a problem even with intel baseline defaults? Really?
"Disabling C-States" <--is that a problem even with intel baseline defaults? Really?
If those are a problem, then the CPUs can not be run at their baseline spec, its using too much power and generating too much heat, there are workloads that will max them out all the time.

PL1 and 2 have been set equal now, with unlimited tau, is that a problem now too?

All the things you list are just additional things, not the cause of the issue but more things you can do to get a cooler system that won't degrade.

– Additional settings which may increase the risk of system instability:
– Disabling C-states
– Using Windows Ultimate Performance mode
– Increasing PL1 and PL2 beyond Intel® recommended limits

The big and main issue intel is stating is this here, this causes all the other things listed below it to make the CPU run way above its limits since the limits are now disabled.

Intel has observed 600/700 Series chipset boards often set BIOS defaults to disable thermal and power delivery safeguards designed to limit processor exposure to sustained periods of high voltage and frequency, for example:

 

[rluker5]

So all of the settings that intel has been OK with, either tacitly or explicitly encouraging motherboard manufacturers to use. Why sell K SKUs and charge a premium for them if they are already at their maximum with intel baseline defaults, and why sell OC chipsets that allow this? Some of these are absurd:
"Using windows ultimate performance mode" <--is that a problem even with intel baseline defaults? Really?
"Disabling C-States" <--is that a problem even with intel baseline defaults? Really?
If those are a problem, then the CPUs can not be run at their baseline spec, its using too much power and generating too much heat, there are workloads that will max them out all the time.

PL1 and 2 have been set equal now, with unlimited tau, is that a problem now too?

What are Intel's baseline defaults?

Here's the ones for my chip for example.
https://ark.intel.com/content/www/u...900kf-processor-36m-cache-up-to-5-80-ghz.html

Intel Core i9-13900KF Specs

Raptor Lake-S, 24 Cores, 32 Threads, 3 GHz, 125 W

www.techpowerup.com

There are some programs you can run that make holding Intel's maximum difficult for some setups, but the maximum isn't guaranteed for all combinations of CPU, mobo, cooling and application. Just like AMD Intel has "up to". if the motherboard forces these frequencies regardless of power or thermal limitations then there may be oveclocking related instabilities. Overclocking performance isn't guaranteed. Some chips can only do the baseline, most can do more and some luckier ones can do much more.

 

[D1v1n3D]

I have to say for competitive reasons it is pretty sad and sucks that Intel keeps fumbling the ball so badly. Releasing the 13th and 14th gen in there state alone is just a disaster waiting to happen, with how hot they run could quite literally cause fires in some situations! But, no one cares to talk about that. Feeding a board with over 300w of power just to get that higher clock that still loses to a much lower clocked CPU that's almost a third the power usage is a tell tail sign Intel doesn't have anything new from core to this hybrid core system still Intel one in the same from core 4th Gen just push the chips to their thermal and frequency limits regardless of risks and when shit hits the fan blame everyone Else.

 

[Digital~Dreams]

Intel issues a state regarding stability issues concerning 13th and 14th Generation CPUs on 600-and 700-series motherboards made by respective manufacturers.

Intel issues statement about CPU crashes, blames motherboard makers — BIOSes disable thermal and power protection, causing issues : Read more

How do these crashes appear on folks systems ?. I've had an intermittent (once every few hours) problem where games will suddenly jump to the desktop (like an ALT-TAB). Nothing helpful ever appears in the windows logs and an app that tracked what changed the window focus showed nothing either !?. I can always then select the game on the taskbar and get right back to where I was but still frustrating.

 

[TerryLaze]

I have to say for competitive reasons it is pretty sad and sucks that Intel keeps fumbling the ball so badly. Releasing the 13th and 14th gen in there state alone is just a disaster waiting to happen, with how hot they run could quite literally cause fires in some situations! But, no one cares to talk about that. Feeding a board with over 300w of power just to get that higher clock that still loses to a much lower clocked CPU that's almost a third the power usage is a tell tail sign Intel doesn't have anything new from core to this hybrid core system still Intel one in the same from core 4th Gen just push the chips to their thermal and frequency limits regardless of risks and when shit hits the fan blame everyone Else.

For competition reasons it's ok, AMD CPUs actually DID catch on fire so intel just crashing is still much more competitive and better for the end user.

View: https://www.youtube.com/watch?v=kiTngvvD5dI&t=221s&ab_channel=GamersNexus

How do these crashes appear on folks systems ?. I've had an intermittent (once every few hours) problem where games will suddenly jump to the desktop (like an ALT-TAB). Nothing helpful ever appears in the windows logs and an app that tracked what changed the window focus showed nothing either !?. I can always then select the game on the taskbar and get right back to where I was but still frustrating.

Far as I know it only happens while unreal engine games compile shaders before the game starts, at least that's the only thing that was reported from any big outlet.

 

[thestryker]

This really comes back to Intel giving the motherboard manufacturers too much leeway. Intel doesn't have strict mandates for many of the settings on their CPUs and the ones with min/max are open to pretty broad interpretation. Intel also gives them tools to help determine settings, but that doesn't mean motherboard manufacturers are tuning these with a board they pull off a retail line.

Buildzoid did a couple of commentary videos based on his experience and the Intel data sheets:

View: https://www.youtube.com/watch?v=8yatSqh5hRA

View: https://www.youtube.com/watch?v=4WNiPQ3PTdE

So their guidance should have been "When we say our recommended specs we really mean those are our recommended specs and not our "recommended specs", so you should actually follow them?" Do explain Mr. Shaikh how Intel should have been more strict in their guidance.

Intel absolutely has the power here and can force the motherboard manufacturers to use their settings as default. Intel hasn't done this much since motherboard manufacturers first started messing with default settings with IVB. Intel obviously isn't going to control other profiles and/or the ability to modify settings, but the default behavior is something they clearly need to.

How do these crashes appear on folks systems ?. I've had an intermittent (once every few hours) problem where games will suddenly jump to the desktop (like an ALT-TAB). Nothing helpful ever appears in the windows logs and an app that tracked what changed the window focus showed nothing either !?. I can always then select the game on the taskbar and get right back to where I was but still frustrating.

This sounds like some sort of software issue. It's definitely not what's happening due to these problems as these are all crash situations where the software or computer are crashing completely.

 

[hotaru251]

So all of the settings that intel has been OK with, either tacitly or explicitly encouraging motherboard manufacturers to use.

wrong.

Intel doesn't accept MB's settings that are out of their defined spec.

They will deny warranties for it even.

The "default" is technically overclocked on most MB's.

That isnt Intels fault.

Why sell K SKUs and charge a premium for them if they are already at their maximum with intel baseline defaults, and why sell OC chipsets that allow this?

as some ppl enjoy OCing and they as a business are in it for profit (so if there is a desire they will have the product) and they specifically don't cover that risk should you do so.

The issue is entirely on MB makers making non stock settings the default.

Intel hasn't done this much since motherboard manufacturers first started messing with default settings with IVB. Intel obviously isn't going to control other profiles and/or the ability to modify settings, but the default behavior is something they clearly need to.

which is a good thing.

Nvidia's control over what people can do w/ their product is a bad thing.

Intel acting like that would instantly remove MB vendors ways to differentiate themselves.
Only thing Intel should be firm about is the default settigns being at their stock values not the MB's overclocked values.

(and intel isnt only one affected as when i built my AMD system I kept having an issue and found out it was due to MB's custom oc on by default and it was causing instability due to trying to run too low of voltage (only undid that it ran fine). Thank god for crash reports being good now-a-days.

 

[CmdrShepard]

Intel doesn't accept MB's settings that are out of their defined spec.

Err, no?

K SKUs will accept any settings.

If Intel really wanted to limit "unlocked" CPUs to always stay within safe operating zone, they could have simply narrowed down the acceptable voltage / power / current / thermal range which the CPU will accept when programmed via MSRs.

They are the ones who allowed this, raked in the money for those "unlocked" CPUs, and are now blaming the mainboard manufacturers for doing what they allowed them to do by removing safeguards from the hardware -- ultimate hypocrisy.

The issue is entirely on MB makers making non stock settings the default.

No it's not -- it's Intel's fault for allowing software to push CPUs into unsafe operating zone by programming its MSRs. You can have BIOS set reasonable defaults and not lock MSRs and then user can install Intel XTU in Windows and set unreasonable values and the result will be the same because the crux of the issue is "unlocked" means "no limits".

If you want to prevent CPU from running at 1.5 V then don't let anyone write the value >= 1.5 V into Vcc control MSR. It's that simple, but no -- Intel wanted that swet money for premium O/C CPUs and now they are washing their hands by blaming others.

 

[JarredWaltonGPU]

Intel hasn't done this much since motherboard manufacturers first started messing with default settings with IVB.

I know from experience that motherboard makers started messing around with FSBs and other settings loooooong before Ivy Bridge, which came out in 2012. For sure there were mobo manufacturers using up to a 103 MHz FSB setting in the Pentium 4 days, and at least in the 101~102 range before that. Pentium III boards running 67~68 MHz instead of 66.67 MHz on the bus also existed.

Back when Tom's Hardware and AnandTech were brand-new sites in the late 90s, there were reviews and roundups that discussed the default bus speeds and how many enthusiast motherboards (esp. Asus and MSI) were goosing the bus and interface speeds slightly just so they'd show up at the top of the performance charts.

The thing is, Intel used to build in much higher margins on its CPUs. Chips sold as "300 MHz" often had at least 20% headroom on basically every chip, so you could run them all at >360 MHz without issue. The word on the street was that binning would identify chips that failed at higher clocks and then rate them for at least 20% lower clocks for retail.

Of course, all the other changes that have happened in the past few decades also play a role. We didn't have multiple cores changing speeds every few microseconds back in the 486 and Pentium days. We just had a static voltage and frequency. Speed Step and all the other tech that dynamically alters clocks, voltages, etc. is great but also adds complexity. This latest generation seems to have much tighter limits, and the mobo vendors just assumed everything was "business as usual."

I'm going to have to go check that MSI video to see what else might be suggested to bring about full stability. I still have occasional quirks pop up, and I can't say for certain whether they're due to the CPU and BIOS settings, or something else.

 

[rluker5]

How do these crashes appear on folks systems ?. I've had an intermittent (once every few hours) problem where games will suddenly jump to the desktop (like an ALT-TAB). Nothing helpful ever appears in the windows logs and an app that tracked what changed the window focus showed nothing either !?. I can always then select the game on the taskbar and get right back to where I was but still frustrating.

That could be a ram thing, a CPU thing, or even some monitoring software conflict thing.
Probably easiest to turn off monitoring software (if you run it all of the time) and see if the problem goes away.

On the CPU side I had a similar problem for a while probably because I like fiddling with settings and running my 13900kf at the lowest possible voltage. I found running at almost the same voltage, but with a higher LLC to make the higher clocks more stable helped. But you will also have to reduce voltage to offset the higher voltage your raised LLC will give you on your Asus board or else see much extra heat. Check these charts I just made on this post: https://forums.tomshardware.com/thr...pdates-and-user-guides.3843249/#post-23250629
Also giving an extra 20mv to 30mv adaptive mode offset to CPU L2 voltage seemed to help all clocks with no noticeable extra heat. I think that is what fixed it for me, but the raising LLC also helps my stability at the top clocks.
And if you are undervolting don't forget to undervolt both the core and cache sections or else it only applies mostly to your top clocks.
And check your IMC VDD in HwInfo, I have DDR5 so this might not apply, but mine was over 100mv too high, which might not be good for stability.
Finally, turning off HT helps stability, temps and usually gaming performance. If you feel like doing that.

View: https://youtu.be/I8DJITHWdaA?t=632


As far as the ram stability question, you know a lot more about what you've done with your DDR4 than I do, but I do know ram tuning can turn into a quagmire so maybe worry about the hardest thing to check last.

 

[thestryker]

I know from experience that motherboard makers started messing around with FSBs and other settings loooooong before Ivy Bridge, which came out in 2012. For sure there were mobo manufacturers using up to a 103 MHz FSB setting in the Pentium 4 days, and at least in the 101~102 range before that. Pentium III boards running 67~68 MHz instead of 66.67 MHz on the bus also existed.

Yeah I meant more the current crop of MCE/power settings as that's when this started.

I rememeber having to deal with the >100 MHz clock settings in the one P4 box I had because the video card I was using was finicky for some reason and just wouldn't run if the AGP clocks were messed with. Bus clocks being decoupled from FSB is one of the best changes I can remember as far as overclocking is concerned.

Of course, all the other changes that have happened in the past few decades also play a role. We didn't have multiple cores changing speeds every few microseconds back in the 486 and Pentium days. We just had a static voltage and frequency. Speed Step and all the other tech that dynamically alters clocks, voltages, etc. is great but also adds complexity. This latest generation seems to have much tighter limits, and the mobo vendors just assumed everything was "business as usual."

I think part of it is also how far outside the boundaries these CPUs can go without dying. If you consider base clock to boost clock the 13900K-14900KS are basically operating like a Celeron 300A did clockspeed wise.

I hope Intel does release some specifics about what's going on because so far it seems like motherboard manufacturers are playing it way too close to instability. I get why they would do this as putting more voltage through the CPU would drop performance when running into a thermal limit.

which is a good thing.

Nvidia's control over what people can do w/ their product is a bad thing.

Intel acting like that would instantly remove MB vendors ways to differentiate themselves.
Only thing Intel should be firm about is the default settigns being at their stock values not the MB's overclocked values.

I can't tell if you misunderstood or are just agreeing because this is exactly what I'm advocating for.

 

[cyrusfox]

As this appears to only be an issue for K type chips, utilized on Z type motherboards, my default expectation is users of said combination that paid for that feature set would learn to utilize those settings to find a stable performance profile. It is unfortunate that today there is not a default baseline or configuration that is easily toggled for the majority of these platforms. It appears Mobo maker has been given a timeline to make that the default by May before Intel publicly releases their findings.

I run a non-k 13900 on a H670 motherboard and I am limited to a 100mV undervolt... With no dynamic tuning (its a static offset). I have it to the minimum -100mV, and the CPU still goes to a much higher default voltage then I would ever set one to even back to my Water OC days (CPU-V-core is normally between 1.38 to 1.43 on no load 5.3GHz turbo when set to high performance mode, balanced will let it bring down multiplier and V-core to sane levels[0.74V]). I have tried playing with some of the other settings available on my motherboard but the options neuter the chips performance (Goes from performing like a 13900k to 60% of the same performance on SC/MC).

I learned my lesson, next time I will get a Z mobo chipset so I can have all the controls present. I Really don't care that Intel sells K sku with cpu OC locked down, as honestly the non-K perform equivalent and the Ks have nearly zero headroom for OC due to the tight binning done to remain competitive with AMD. What is annoying is the difference in Motherboard chipsets (Z vs H vs B), the default settings while good enough for someone elses PC, not enough for me. I personally only get the K chip when I can't bother to wait 6 months for the non-k release. Excited for arrowlake, and so far I have not side graded for a 14900, although I am very tempted as I have a few nephews that would love the upgrade from there 12600k's I originally bought them 2 years back.

 

[wingfinger]

A long time ago you could reset to a default configuration or to an optimized "default" configuration. Of course, today, no.

There used to be an intel bios compliance test. It could be BIOS Implementation Test Suite. There seems to be an open source that hasn't been updated (biosbits). Even when it was current, reviewers didn't post the compliance.

 

[bit_user]

Intel requests system and motherboard manufacturers to provide end users with a default BIOS profile that matches Intel recommended settings.

Intel strongly recommends customer’s default BIOS settings should ensure operation within Intel’s recommended settings.

@JarredWaltonGPU & @PaulAlcorn , I'd love to see Toms re-test the i9-14900K with these defaults to see how much performance is lost relative to the original i9-14900K review:

https://www.tomshardware.com/news/intel-core-i9-14900k-cpu-review
​

 

[-Fran-]

In yet another "I'm shocked! SHOCKED! ... Well, not that shocked" moment, Intel blames anyone else but themselves and the sheeple agrees.

Both Intel and AMD are also to blame if motherboard vendors run away doing wacky stuff. But well, what am I even saying. AMD is only to blame, because it's AMD, but not when it's Intel.

Sarcasm aside, I wonder what will happen now... Oh, I know: nothing.

Regards.

 

[ingtar33]

Intel doesn't accept MB's settings that are out of their defined spec.

They will deny warranties for it even.

The "default" is technically overclocked on most MB's.

That isnt Intels fault.

Intel has been endorcing MB doing this for almost 15 years now. this is like the costa concordia owners claiming they never allowed "sail by salutes" after it sank hitting a rock while doing one. That was a lie by the way, they never forbit it and in fact endorced the practice. much like intel has for 15 years now with MB venders.

AMD used to be all for mb venders doing the same for thier chips till AM4 and ryzen, after the early issues they had with the early gen ryzen chips burning out. they sorta put their foot down with the mb makers about the recommended specs not being a "guideline" but actually the rated specs.

 

[watzupken]

I believe that Intel chose to be "grey" with regards to the guidelines to the motherboard manufacturers for this reason. If nothing happens, the "unlocked" over the recommendation power limit will make their CPU perform better as long as it is not crashing. When something goes wrong and becomes more widespread, Intel can just turn around and claim its the motherboard manufacturer's problem. In reality, both are at fault because its clear that Intel pushed their chips too hard, and motherboard manufacturers are pushing for more relaxed power limits.

 

[Amdlova]

months ago intel removes "CEP" some cpus gain almost 10% of "free performance".

disable CEP (CurrentExcursion Power).

Some z790 get that but they want install in all motherboards...

Remove the limits to fry some cpu!

The 35w cpu changed to 50w almost gives 20% off boost with that CEP maybe achives 30% lol
It's free why not :S

 

[TerryLaze]

No it's not -- it's Intel's fault for allowing software to push CPUs into unsafe operating zone by programming its MSRs. You can have BIOS set reasonable defaults and not lock MSRs and then user can install Intel XTU in Windows and set unreasonable values and the result will be the same because the crux of the issue is "unlocked" means "no limits".

If you want to prevent CPU from running at 1.5 V then don't let anyone write the value >= 1.5 V into Vcc control MSR. It's that simple, but no -- Intel wanted that swet money for premium O/C CPUs and now they are washing their hands by blaming others.

Intel sells those as non-k CPUs ,they even sell T versions with even lower settings, if clocks are locked you can have everything else (power volts amps) at unlimited and the cpu will always be 100% stable and safe because there is only that much that it needs to draw for the clocks it can reach.

If you as the customer wants a CPU that can do better than that then you have to also live with the mess that comes with overclocking.

 

[phitinh81]

For competition reasons it's ok, AMD CPUs actually DID catch on fire so intel just crashing is still much more competitive and better for the end user.

View: https://www.youtube.com/watch?v=kiTngvvD5dI&t=221s&ab_channel=GamersNexus


Far as I know it only happens while unreal engine games compile shaders before the game starts, at least that's the only thing that was reported from any big outlet.

"Intel just crashing is still much more competitive & better for the end user" Is this English ??? That Gamernexus video just showed CPU can burn by pumping SOC voltage out of spec. It doesn't mean anything because any electronic circuit can be smoked that way. If you're not a bot, your reasoning is adorable i must admit

 

[phitinh81]

This really comes back to Intel giving the motherboard manufacturers too much leeway. Intel doesn't have strict mandates for many of the settings on their CPUs and the ones with min/max are open to pretty broad interpretation. Intel also gives them tools to help determine settings, but that doesn't mean motherboard manufacturers are tuning these with a board they pull off a retail line.

Buildzoid did a couple of commentary videos based on his experience and the Intel data sheets:

View: https://www.youtube.com/watch?v=8yatSqh5hRA

View: https://www.youtube.com/watch?v=4WNiPQ3PTdE


Intel absolutely has the power here and can force the motherboard manufacturers to use their settings as default. Intel hasn't done this much since motherboard manufacturers first started messing with default settings with IVB. Intel obviously isn't going to control other profiles and/or the ability to modify settings, but the default behavior is something they clearly need to.

This sounds like some sort of software issue. It's definitely not what's happening due to these problems as these are all crash situations where the software or computer are crashing completely.

Intel were asking for trouble in the first place. They know what perfectly stable specs should be. But they chose to give leeways to motherboard vendors, boosting performance and remain competitive. If anything happen down the road, they can blame MB vendors as they just did. If any vendor doesn't crank up specs, their boards will just collect dust on shelf. Business is brutal these days.