Intel Statement on Stability Issues: "Motherboard Makers to Blame"

[Crackong]

No CPUs have been blown up that we know of, this isn't the "Meltdown" fiasco, there has been some instability in certain workloads.

If CPUs were blown up people gets RMA and case would move a lot quicker.
What leaves everyone's mouth bitter is the CPUs aren't blowing up, no RMA, but confirmed decreased performance, and no clear solution is provided, yet.

I still think there is a lot of jumping on this topic to attack Intel, and not enough people criticising the fact that board partners who should know better are possibly comically incompetent to the point of not being able to copy and paste several numbers from a datasheet, or potentially still trying to gain competitive advantage by using the wrong values.

It is Intel's CPU, it is their job to make sure the motherboard vendors having a correct 'Default' profile so it works 100% of the time.
This lack of communication alone is a big issue and is one of the Intel biggest fault.

And, if your CPUs are this fragile, measurements should be taken to 'prevent' the partners further messing it up.
Like AMD, with their X3D voltage issue, they forced new voltage setting very quickly and RMA every affected case.
Like Nvidia, Nvidia does a great job make sure the AIB cannot mess up their GPUs and, if something's up like the 12vhpwr issue, Nvida took the responsibility and took care every affected case.

Please noted in the above mentioned cases,
Although customers do blame the AIB partners,
But AMD/Nvidia themselves didn't actively placed the blame on their partners.
They just went in, solved the problem, and get out ASAP.

if they can do it, why not Intel ?

 

[BoggledBeagle]

BTW the fact itself that Intel acknowledged the problem and is preparing something is a proof, that they are getting a lot of degraded / unstable chips back. They would happily ignore the problem if it was not biting them badly.

 

[londiste]

Are you sure?

Since 14900KS had a PL1/PL2 = 150/320, which is differ from regular 14900K's 125/253
If they had the same baseline profile, it will render them basically the same SKU.

This is a bit of spec bullshit from Intel. ARK page for CPUs has Processor Base Power which is not PL1. Maximum Turbo Power is reasonably enough PL2.
The spec values for PL1/PL2 are in the Datasheet.
14900K Extreme profile and 14900K non-Extreme are the same. Extreme profile is not and should not be applicable for Base Profile.

The fact that Intel's CPU power delivery specification is so convoluted, with so many knobs and dials, would reasonably suggest a pressing need for Intel to carefully validate any firmware that board partners release, in order to prevent blown up CPUs.

Oh sweet summer child. What makes you think only Intel's power delivery has many knobs and dials?

 

[Dr. Dro]

BTW the fact itself that Intel acknowledged the problem and is preparing something is a proof, that they are getting a lot of degraded / unstable chips back. They would happily ignore the problem if it was not biting them badly.

The 13900K itself is barely over a year old. I have owned my 13900KS for a little over 370 days myself, I find it rather unlikely that these chips would degrade so fast, unless subjected to 110+C and 350+W constantly and even then. Sensationalist "news" articles tend to cause panic and exacerbate situations, which tend to reflect on general consumer mood and propensity to return "faulty" systems.

I have good reason to believe that I was affected by said stability problems (at least with the last 1F BIOS, and there's some likelihood it was my fault as well), since updating to the 1G BIOS I made sure I triple checked every single one of my settings and memory subtimings, and I have not experienced any BSOD's since. Something funny though, whenever my computer was about to crash, the scrambled graphics bug on the Nvidia drivers that affected Chromium would trigger something fierce.

It is Intel's CPU, it is their job to make sure the motherboard vendors having a correct 'Default' profile so it works 100% of the time.
This lack of communication alone is a big issue and is one of the Intel biggest fault.

And, if your CPUs are this fragile, measurements should be taken to 'prevent' the partners further messing it up.
Like AMD, with their X3D voltage issue, they forced new voltage setting very quickly and RMA every affected case.
Like Nvidia, Nvidia does a great job make sure the AIB cannot mess up their GPUs and, if something's up like the 12vhpwr issue, Nvida took the responsibility and took care every affected case.

Please noted in the above mentioned cases,
Although customers do blame the AIB partners,
But AMD/Nvidia themselves didn't actively placed the blame on their partners.
They just went in, solved the problem, and get out ASAP.

if they can do it, why not Intel ?

Ultimately this is my sole problem with Intel's response. They should be far more proactive and shield themselves less from any potential blame. But then again, if you look at who's running Intel's PR at the moment, you're going to understand this stance. It's the same one AMD took until very, very recently

 

[Robin Seina]

Compare current statement with Intel interview with Anandtech in 2019:

Ian Cutress: One of the things we’ve seen with the parts that we review is that we’re taking consumer or workstation level motherboards from the likes of ASUS, ASRock, and such, and they are implementing their own values for that PL2 limit and also the turbo window – they might be pushing these values up until the maximum they can go, such as a (maximum) limit of 999 W for 4096 seconds. From your opinion, does this distort how we do reviews because it necessarily means that they are running out of Intel defined spec?


Guy Therien: 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.
We’re going to be very crisp in our definition of what the difference between in-spec and out-of-spec is. There is an overclocking 'bit'/flag on our processors. Any change that requires you to set that overclocking bit to enable overclocking is considered out-of-spec operation. So if the motherboard manufacturer leaves a processor with its regular turbo values, but states that the power limit is 999W, that does not require a change in the overclocking bit, so it is in-spec.

Source: https://www.anandtech.com/show/1458...ng-an-interview-with-intel-fellow-guy-therien

 

[londiste]

The 13900K itself is barely over a year old. I have owned my 13900KS for a little over 370 days myself, I find it rather unlikely that these chips would degrade so fast, unless subjected to 110+C and 350+W constantly and even then. Sensationalist "news" articles tend to cause panic and exacerbate situations, which tend to reflect on general consumer mood and propensity to return "faulty" systems.

I do not think the chips degrade. There is a (relatively) new use case that is very picky about CPU stability. I got the same errors - like running out of VRAM in Unreal Engine games when there was clearly enough VRAM - but on a 7800X3D. Took a moment and the news about these being caused by CPUs before I went and changed the Curve Optimizer to a bit less than -30 on all cores. Nothing else before this had been failing or even given any indications of failing.

 

[dgianstefani]

Compare current statement with Intel interview with Anandtech in 2019:

Ian Cutress: One of the things we’ve seen with the parts that we review is that we’re taking consumer or workstation level motherboards from the likes of ASUS, ASRock, and such, and they are implementing their own values for that PL2 limit and also the turbo window – they might be pushing these values up until the maximum they can go, such as a (maximum) limit of 999 W for 4096 seconds. From your opinion, does this distort how we do reviews because it necessarily means that they are running out of Intel defined spec?


Guy Therien: 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.
We’re going to be very crisp in our definition of what the difference between in-spec and out-of-spec is. There is an overclocking 'bit'/flag on our processors. Any change that requires you to set that overclocking bit to enable overclocking is considered out-of-spec operation. So if the motherboard manufacturer leaves a processor with its regular turbo values, but states that the power limit is 999W, that does not require a change in the overclocking bit, so it is in-spec.

Source: https://www.anandtech.com/show/1458...ng-an-interview-with-intel-fellow-guy-therien

If all mobo makers did was change PL values, the CPU would still have all of the safeguards and algorithms to ensure stability.

The issue is that the board partners also disable many of the Intel boost algorithms, C states etc.

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:

• Disabling Current Excursion Protection (CEP)
• Enabling the IccMax Unlimited bit
• Disabling Thermal Velocity Boost (TVB) and/or Enhanced Thermal Velocity Boost (eTVB)
• 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

This one change doesn't change the voltage curve much, i've tested it on many builds I have done, hence falling under the "additional settings that may increase the risk" category.
These settings are quite significant -

• Disabling Current Excursion Protection (CEP)
• Enabling the IccMax Unlimited bit
• Disabling Thermal Velocity Boost (TVB) and/or Enhanced Thermal Velocity Boost (eTVB)

The crashes are from voltage going too high or too low. This can also be exacerbated by VRMs failing to keep up with boost clock changes, and over/undershooting voltage targets.

Modifying the target frequency (MCE all core boost etc and other names for this), voltage LLC, and other settings will adjust the voltage curve outside of spec.

 

[Robin Seina]

If all mobo makers did was change PL values, the CPU would still have all of the safeguards and algorithms to ensure stability.

The issue is that the board partners also disable many of the Intel boost algorithms, C states etc.


This one change doesn't change the voltage curve much, i've tested it on many builds I have done, hence falling under the "additional settings that may increase the risk" category.
These settings are quite significant -

• Disabling Current Excursion Protection (CEP)
• Enabling the IccMax Unlimited bit
• Disabling Thermal Velocity Boost (TVB) and/or Enhanced Thermal Velocity Boost (eTVB)

The crashes are from voltage going too high or too low. This can also be exacerbated by VRMs failing to keep up with boost clock changes, and over/undershooting voltage targets.

Well, Intel certainly knew about and encouraged this MB vendor behaviour for at least last 6 years. It gave him competitive and marketing advantage, after all. Now, the company is shifting all the blame on said MB vendors.

 

[chrcoluk]

Part 2 of the debacle, Asus baseline.

 

[Crackong]

Part 2 of the debacle, Asus baseline.

Thanks, watching right now.

2 minutes in and Buildzoid already shown something interesting, the optimized default of ASUS APEX board BIOS 1202 was PL1/PL2 = 253/4095
That's is like..insane and it consumes 360W in R15 and crash R15

The baseline default on the APEX seems a lot resonable than Gigabyte, Buildzoid's 14900k runs 10% faster than the Gigabyte's baseline default.

 

[chrcoluk]

So summary of my thoughts based on both documents.

i9 CPUs listed as baseline power of 150w (which seems to be the KS models), have 253/253 as perf spec. Extreme config of 320/320.
8+8 and 8+16 models which have 125w listed as base power have 253/125 as perf spec. But also have an extreme config as 253/253.
Baseline spec seems to be 188/125.

Document here, originally posted by @dgianstefani

13th Generation Intel® Core™ and Intel® Core™ 14th Generation Processors Datasheet, Volume 1 of 2

WW14.3 Metadata Changes only This describes all publicly disclosed specifications including electrical characteristics, mechanical, and component functionality, a list of major features, a functional description, and an architectural overview.

www.intel.com

 

[mkppo]

They do just say it.

Have a look at the datasheet.

What other companies do with Intel products is up to them.

If I buy a car and tune the engine until it explodes, is this the fault of the car manufacturer?

No, but if the car manufacturer says this car tuned to 500BHP with a stage one turbo kit is "within spec" and later dies slowly it's absolutely on the manufacturer. That's exactly what happened here, as evident by the response they provided to Ian Cuttress when talking about consumer and professional boards (not just workstation mind you, please read it carefully or see the clip).

Good on Hardware Unboxed to lay it down on intel and they absolutely deserve it. It's a fact that they knew exactly what the board makers were doing, and did nothing about it but rather encouraged the behaviour. I remember some of the senior reviewers a few years back were pretty confused about the whole power limits thing, the ambiguity and lack of action on intel's part. I know for a fact Ian was, and there were some podcasts where they spoke about it in more detail and people even reached out to intel but they had absolutely zero issues with board partners and everything was 'within spec'.

You can't really defend intel here and just say "oh they should have enforced the board manufacturers more that's their only fault". That's really not the only fault there is it. When people started commenting on other issues, you admitted that the PL1=PL2 was not defined properly earlier and recently resurfaced. There are many more cracks in Intel's spec, and lots of ambiguity and looseness in their 'guide' which is an abhorrent mess. Most of it is covered by HWU.

Just FYI, there's no speculation on Steve's part and I 100% agree with him.

 

[chrcoluk]

Thanks, watching right now.

2 minutes in and Buildzoid already shown something interesting, the optimized default of ASUS APEX board BIOS 1202 was PL1/PL2 = 253/4095
That's is like..insane and it consumes 360W in R15 and crash R15

The baseline default on the APEX seems a lot resonable than Gigabyte, Buildzoid's 14900k runs 10% faster than the Gigabyte's baseline default.

Yep its a mess.

He also raises a point which is bad news, previously if you had a lottery loser chip that wasnt stable out of the box you would probably get a new one with RMA, now they might refuse swap if its stable on baseline.

 

[Darmok N Jalad]

I see it rather simply. Intel should have a safe baseline as the platform default, and board partners should be held to that as the first boot default, or no license. If board partners want to run wild on power specs, fine, but issue the appropriate warning before it gets enabled. It’s what any system builder should expect, honestly.

I simply don’t believe that Intel didn’t know this wasn’t happening all along. They sure seem to have opened the door for this and just left themselves an out for when this finally came home to roost. They got a few generations of better-looking benchmarks out of the deal.

Intel just needs to take ownership of not enforcing proper defaults, but that will only cost them in the long run if they continue down their current design path of allowing for insane power limits. I remember when this all started, where Intel CPUs lacked in performance at default settings, but took off once you unlocked the power limits. It seems it only took one generation for those defaults to start getting ignored.

 

[Crackong]

now they might refuse swap if its stable on baseline.

Adding insult to injury, even if a user find it unstable with baseline, it could be on the difference between vendor's baseline profiles.
RMA the CPU and the tester test it on a Gigabyte.....
Now that user get stuck with a 'No problem' CPU which doesn't work on his/her MB on hand...

 

[asdkj1740]

1. intel document does state there are two profiles for 13900k/13900kf/14900k/14900kf which is rated at tdp 125w.
2. the profiles in the doc are,
2a. pl1=125w & pl2=253w & iccmax=307a & ac_ll=1.1omhm
2b. pl1=253w=pl2 & iccmax=400a & ac_ll=1.1omhm (this is called extreme profile)
so both files are spec by intel, both are OK to be used.
3. actually the same intel doc got lots of rev, and in the rev002 which was released in late 2022, for the same tdp=125w cpu like 13900k, the pl1 was 125w and the pl2 was 188w.
so, three profiles created by intel, but 125 188 one seemed to be cancelled.
4. what igor leaked in 2021 about raptor lake s 125w cpu, that spec turns out to be almost the same as what gigabyte now has implemented (pl1=125w & pl2=188w & iccmax=249a & ac_ll=1.7)
5. what igor leaked in 2021 about raptor lake s 125w cpu, that spec was called "baseline"
6. if you search "baseline" in the latest intel doc, there is no such result that is related to power/voltages/current.
7. why gigabyte set things like that in the baseline profile, maybe the engineer was told to make a intle baseline profile but he didn't know what the fuck was that, so he looked up intel latest doc but found nothing about BASELINE, then he had to look up even earlier, at last he did find out intel baseline, which was the same as what igor leaked back in 2021, so gigabyte copied that and tested that and found out 200a iccmax was way too low, then gigabyte eventually decided to increase the iccmax by themself.

https://twitter.com/i/web/status/1784864270573851056

 

[R0H1T]

Ultimately this is my sole problem with Intel's response. They should be far more proactive and shield themselves less from any potential blame. But then again, if you look at who's running Intel's PR at the moment, you're going to understand this stance. It's the same one AMD took until very, very recently

I mean I can't stress this enough ~ Intel has pushed BIOS updates & ucode updates blocking free OCing over at least 5 gen of motherboards, going as far as a year(?) down the line after the products were sold to block them off! Anyone defending Intel over this is either totally ignorant of this fact or just (short of) a paid shill

Let me repeat in case it's not clear ~ Intel can force their "board partners" to adhere to their specs in a second, if they wanted to! No pipsqueak would try going against that after Intel's forced their hand.

And we all know the reason they didn't till now

 

[Tek-Check]

The processors are stable if configured as advertised (PL1=PL2), the rest to spec.
Everything else is mobo maker deviation.

It's way more messy. Massive power unlocking is within the spec, according to what Intel officially said in interview with Dr Cutress and according to additiona information on their website, one click away under 253W.

They consider OC to be changes in the multiplier. So, if motherboard vendors keep the multiplier intact, technically a CPU operates within the spec, no matter how much power they throw at it. This utter mess is a sole responsibility of Intel not willing to define clear power boundaries.

Intel clearly reads on their website that 253W is not set in stone as Maximum Turbo Power can be configurable by OEM.
"The maximum sustained (>1s) power dissipation of the processor as limited by current and/or temperature controls. Instantaneous power may exceed Maximum Turbo Power for short durations (<=10ms). Note: Maximum Turbo Power is configurable by system vendor and can be system specific."

Intel® Core™ i9 processor 14900K (36M Cache, up to 6.00 GHz) - Product Specifications | Intel

Intel® Core™ i9 processor 14900K (36M Cache, up to 6.00 GHz) quick reference with specifications, features, and technologies.

www.intel.com

 

[Solid State Brain]

They consider OC to be changes in the multiplier. So, if motherboard vendors keep the multiplier intact, technically a CPU operates within the spec, no matter how much power they throw at it. This utter mess is a sole responsibiluty of Intel not willing to define clear power boundaries.

The specs are more or less clear.
- Sustained TCase (IHS temperature) must be < TCaseMax, which for 13/14-gen 125W processors is 61.9 °C.
- TJunction must be always < TJmax (100 °C)
- TCase has a certain thermal inertia. The processor is allowed to exceed base power as long as TCase remains below the spec value. PL2 is intended to take advantage of that thermal inertia to provide a short-term performance boost.
- If the cooling is good enough that Tcase is always below TCaseMax, then PL1 can be increased, or even made equal to PL2.
- The official specs are validated for Intel's own standardized "thermal solution". The minimum spec is that CPUs must be able to sustain base power (e.g. 125W) indefinitely without exceeding TCaseMax.

 

[Dr. Dro]

Intel clearly reads on their website that 253W is not set in stone as Maximum Turbo Power can be configurable by OEM.
"The maximum sustained (>1s) power dissipation of the processor as limited by current and/or temperature controls. Instantaneous power may exceed Maximum Turbo Power for short durations (<=10ms). Note: Maximum Turbo Power is configurable by system vendor and can be system specific."

Intel® Core™ i9 processor 14900K (36M Cache, up to 6.00 GHz) - Product Specifications | Intel

Intel® Core™ i9 processor 14900K (36M Cache, up to 6.00 GHz) quick reference with specifications, features, and technologies.

www.intel.com

That just means it's unlocked. And there goes with the "sensationalism" i was going on about. Guess even Hardware Unboxed needs their views every now and then.

 

[remixedcat]

I mean I can't stress this enough ~ Intel has pushed BIOS updates & ucode updates blocking free OCing over at least 5 gen of motherboards, going as far as a year(?) down the line after the products were sold to block them off! Anyone defending Intel over this is either totally ignorant of this fact or just (short of) a paid shill

Let me repeat in case it's not clear ~ Intel can force their "board partners" to adhere to their specs in a second, if they wanted to! No pipsqueak would try going against that after Intel's forced their hand.

And we all know the reason they didn't till now

didn't they also block under/overvolting on core ultra?? It seems they are, indeed getting more restrictive. You need that for the thin n light laptops.

 

[Steevo]

So, all that sweet sweet 1-3% on some games for 3X the wattage was out of spec? How very sus of Intel, imagine how much heat AMD would get for BIOS issues.....


I wonder if every review will need updated to reflect actual chips that thousands of consumers got that wouldn't run out of the box.

 

[Tek-Check]

That just means it's unlocked. And there goes with the "sensationalism" i was going on about. Guess even Hardware Unboxed needs their views every now and then.

It's interesting, isn't it. Intel allows OEMs to run unlocked profiles, everything is good when benchmarks are high and then blames them in an official preliminary statement when CPUs degrade over time and cause instability.

Being complicit in OEM practices during good times and then slapping them when push gets to shove sounds like washing hands, no?

 

[dir_d]

Clearly there is a problem if there is 6 pages of knowledgeable people bickering about what is the "Baseline". All we can do is wait for Intel's "Official" document regarding this in May. It is fun to try to figure this out ourselves though.

 

[matar]

I am glad I did not go this route and just a month ago i was playing to buy a z790 with a 13900kf for a great deal but i said no i will pass because of the heating issues and then i will need a better AIO and a better case so wasn't worth it so i am waiting on the new 15gen and maybe till 16gen will see i am very happy with my i9-10900KF 24/7 @5.1GHZ all core and Ring @4.6ghz @1.28v