Question DEGRADING Raptor lake CPUs

[Kocicak]

I noticed some reports about degrading i9 13900K and KF processors.

I experienced this problem myself, when I ran it at 6 GHz, light load (3 threads of Cinebench), at acceptable temperature and non extreme voltage. After only few minutes it crashed, and then it could not run even at stock setting without bumping the voltage a bit.

I was thinking about the cause for this and I believe the problem is, that people do not appreciate, how high these frequencies are and that the real comfortable frequency limit of these CPUs is probably at something like 5500 or 5600 MHz. These CPUs are made on a same process (possibly improved somehow) on which Alder lake CPUs were made. See the frequencies 12900KS runs at. The frequency improvement of the new process tweak may not be so high as some people presume.

Those 13900K CPUs are probably highly binned to be able to find those which contain some cores which can reliably run at 5800 MHz. Some of the 13900K probably have little/no OC reserve left and pushing them will cause them to degrade/break.

The conclusion for me is that the best you can do to your 13900K or 13900KF is to disable the 5800 MHz peak, which will allow you to offset the voltage lower, and then set all core maximal frequency to some comfortable level, I guess the maximum level could be 5600 MHz. With lowered voltage this frequency should be gentler to the processor than running it at original 5500 MHz at higher voltage. You can also run it at lower frequencies, allowing for even higher voltage drop, but then the CPU is slowly loosing its sense (unless you want some high efficiency CPU intended for heavy multithread loads).

Running it with some power consumption limit dependent on your cooling solution to keep the CPU at sensible temperature will help too for sure.

 

[igor_kavinski]

At least my 14900K sample does not exceed such voltage at 5/4 GHz.
At 4,5/3,6 GHz is runs at 1,1V, most of the time a bit less than that.

Here are the Geekbench 6 scores for above limits:

Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX with an Intel Core i9-14900K processor.

browser.geekbench.com

Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX with an Intel Core i9-14900K processor.

browser.geekbench.com

Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX vs Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX - Geekbench

Interesting how at ~1.2V, it is faster in just about everything except HDR ST where it nosedives hard, probably due to thermal throttling.

Can you do a test? Compress/decompress any 10GB file a few times at 5/4 GHz setting and see if your CPU passes the test consistently. You can use WinRAR or 7-zip.

 

[Kocicak]

Interesting how at ~1.2V, it is faster in just about everything except HDR ST where it nosedives hard, probably due to thermal throttling.

Can you do a test? Compress/decompress any 10GB file a few times at 5/4 GHz setting and see if your CPU passes the test consistently. You can use WinRAR or 7-zip.

Oh, I have not noticed that, perhaps something interfered with the benchmark, some other application? Or could it be some insta- NO!!!!!!

If I did compression test, how could I compare if the file is still the same as in the beginning?

 

[Hulk]

Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX vs Gigabyte Technology Co., Ltd. Z790 AORUS ELITE AX - Geekbench

Interesting how at ~1.2V, it is faster in just about everything except HDR ST where it nosedives hard, probably due to thermal throttling.

Can you do a test? Compress/decompress any 10GB file a few times at 5/4 GHz setting and see if your CPU passes the test consistently. You can use WinRAR or 7-zip.

Compression/decompression was one of the tests I used when finding day-to-day settings for my 14900K. It is sensitive to frequency/voltage settings.

My hypothesis. Intel binned these parts too aggressively, the guard band was too small. Looks like AMD might have been concerned about the same issue with Zen 5 and they got a little nervous because of what is going on with Intel and decided to "rebin" their parts a little more conservatively.

 

[coercitiv]

Looks like AMD might have been concerned about the same issue with Zen 5

How did you come up with this conclusion?

 

[Hulk]

How did you come up with this conclusion?

Ahead of the launch, AMD found that “the initial production units that were shipped to our channel partners did not meet our full quality expectations."

Obviously they are not revising the architecture in two weeks so it makes sense that this is a binning/quality issue.

I think AMD did some additional quality testing in light of the ongoing Intel fiasco and reflected, "hey there are some niche situations where we might have similar problems." They are recalling the parts and re-binning to assure quality. Good on AMD.

 

[igor_kavinski]

If I did compression test, how could I compare if the file is still the same as in the beginning?

WinRAR/7-zip calculate CRC32 checksum of the file to avoid data integrity issues. During subsequent compression/decompression cycles, the various parts of the die should heat up real good and that should expose the failed TLB or ring bus issue and you should see a checksum failed error message with decompression getting aborted. If not, you are good.

 

[DAPUNISHER]

Ahead of the launch, AMD found that “the initial production units that were shipped to our channel partners did not meet our full quality expectations."

Obviously they are not revising the architecture in two weeks so it makes sense that this is a binning/quality issue.

I think AMD did some additional quality testing in light of the ongoing Intel fiasco and reflected, "hey there are some niche situations where we might have similar problems." They are recalling the parts and re-binning to assure quality. Good on AMD.

I am going full tinfoil brigade and espousing that it's pure gamesmanship.

For good or bad Intel hijacked the news cycle shortly before AMD's next gen launch. Taking back the headlines is the first counter.

Intel delaying the patch until mid August gets countered with "Cool cool, we can wait for you."

And the cherry on top "That's known as a recall. In case you folks over at INTC were wondering what a recall is. Seems like you weren't certain. You're welcome🖕"

 

[samboy]

Intel's upcoming patch will not help Raptor CPU's that have already been impacted & this really falls into two categories:-

1. CPU has already been experiencing sporadic crashes like reported
2. CPU is impacted and degraded; but sporadic crashes have not been observed
- I expect a lot of end users will have this concern. August 15th patch can only do so much

Intel doesn't need to wait until mid August to make a public statement on how they plan to support folks on the above & why haven't they got ahead of this this given their "customer support motto"?

1. Looks like a clear cut product recall (they have admitted fault with their erroneous voltage algorithm)
- Shouldn't the details of this be communicated now?
2. More difficult situation....... would be nice if Intel could provide a tool to measure degradation (possible to do so?) and issue policy statement
- Better for users to be able to schedule their own replacement rather than wait for the CPU to do this for you

Why has Intel been silent on this front?

 

[ikjadoon]

There is no fix for Intel’s crashing 13th and 14th Gen CPUs — any damage is permanent

We got some answers from Intel, and more are on the way.

www.theverge.com

Ugly to disgusting. This needs a proper watchdog / regulatory investigation at this point.

• Intel confirms ANY 65W or higher 13th/14th CPU is susceptible to accelerated degradation.
• Intel will not recall any CPUs.
• Intel will not provide any manufacturing dates / serial number ranges for via oxidation.
• Intel (obviously) confirms all damage is permanent.

[shameless copy-paste of my comment on another forum]

 

[Hitman928]

Why has Intel been silent on this front?

I think an easy explanation is that to admit fault and initiate any kind of mass recall would be extremely painful for Intel both financially and for their reputation, as there are potentially billions of dollars worth of effected product in the market.

 

[samboy]

I think an easy explanation is that to admit fault and initiate any kind of mass recall would be extremely painful for Intel both financially and for their reputation, as there are potentially billions of dollars worth of effected product in the market.

Yes, but that's water over the dam now......... how does Intel think this is going to play out now?

They can get ahead of the issue and try and keep some customer loyalty or get dragged over the coals which will cost them even more money (and unfortunately any class action lawsuit would benefit the lawyers more than customers)

 

[Kocicak]

... any kind of mass recall would be extremely painful for Intel both financially and for their reputation, as there are potentially billions of dollars worth of effected product in the market.

Is there any way to avoid that?

I remember Pat G. boasting about high frequency and how engineers thought such frequencies were impossible. Well they were really impossible and those frequencies were brute-forced by the management, no matter what. They are responsible for the damage.

 

[Hitman928]

Yes, but that's water over the dam now......... how does Intel think this is going to play out now?

They can get ahead of the issue and try and keep some customer loyalty or get dragged over the coals which will cost them even more money (and unfortunately any class action lawsuit would benefit the lawyers more than customers)

Is there any way to avoid that?

I remember Pat G. boasting about high frequency and how engineers thought such frequencies were impossible. Well they were really impossible and those frequencies were brute-forced by the management, no matter what. They are responsible for the damage.

Well, if Buildzoid's speculation is correct, Intel plans on dropping the voltages just enough to slow down the degradation so that most CPUs can at least get through their warranty period before becoming unstable. Then Intel wouldn't have to mass replace CPUs and they can just hope that people will move on as their next gen CPUs are released and fill up the news cycle. They'll take a major hit to their reputation, but a recall would get mainstream news coverage and probably hurt even more. They'll work out deals with their big OEM type customers on the side and the sales reps will take them out to dinner to an extra fancy restaurant that year and then business as usual. Stuff like that has worked for them in the past anyway. (Edit: This very well could come back to bite them, but [hard speculation based on Buildzoid's speculation] it could be how they hope it plays out at this point.)

Edit: @Kocicak added a quote to your post here since I had just finished my reply to samboy when I saw yours.

 

[Gideon]

I think an easy explanation is that to admit fault and initiate any kind of mass recall would be extremely painful for Intel both financially and for their reputation, as there are potentially billions of dollars worth of effected product in the market.

Intel really is in a bind

They can't really do a recall, even if they wanted to, as they don't have anything stable to replace the chips with.

12900K's wouldn't cut it and new 14xxxK chips would still eventually fail (and take 6+ months to produce to replace in that volume)

I guess they only thing they could realistically do is to reimbuse the CPUs. But i can't imagine the cost of that

 

[samboy]

Well, if Buildzoid's speculation is correct, Intel plans on dropping the voltages just enough to slow down the degradation so that most CPUs can at least get through their warranty period before becoming unstable. Then Intel wouldn't have to mass replace CPUs and they can just hope that people will move on as their next gen CPUs are released and fill up the news cycle. They'll take a major hit to their reputation, but a recall would get mainstream news coverage and probably hurt even more. They'll work out deals with their big OEM type customers on the side and the sales reps will take them out to dinner to an extra fancy restaurant that year and then business as usual. Stuff like that has worked for them in the past anyway. (Edit: This very well could come back to bite them, but [hard speculation based on Buildzoid's speculation] it could be how they hope it plays out at this point.)

Edit: @Kocicak added a quote to your post here since I had just finished my reply to samboy when I saw yours.

You're probably right..... their behavior is consistent with trying to "bury" the problem. I've seen this happen when I returned a faulty product (many other folks in the same boat) and manufacturer sent me a check for the cost; with the provision that I keep everything confidential. It was a well known brand and I was surprised that this stayed "buried"; first hand experience that this strategy can work.

My expectation is that this is too big to "bury" now; but point taken that Intel is probably biding their time to test the waters so to speak. You are spot on that this will still take a toll on their reputation either way.

Now, should I short Intel and try and make some money out of this?

 

[Kocicak]

They can't really do a recall, even if they wanted to, as they don't have anything stable to replace the chips with.

They can provide slower CPUs. The question is if consumers would accept such CPUs and what reimbursement for the decline in performance would they require.

 

[Kocicak]

I guess they only thing they could realistically do is to reimbuse the CPUs. But i can't imagine the cost of that

These CPUs are most important parts of systems, would Intel need to cover the cost of a whole system, when without a CPU it is unusable?

 

[samboy]

That can provide slower CPUs. The question is if consumers would accept such CPUs and what reimbursement for the decline in performance would they require.

If their current statements are correct (and remember this is the second or third round now?) then they can replace with new Raptor CPU's with the provision that BIOS be upgraded to mid August release (which means September time frame before really available).

Intel is probably ok if the performance is lower; as long as the clock speeds have not regressed. They typically don't guarantee any benchmark results. Not that this wouldn't be misleading and annoying to a lot of folks

 

[Kocicak]

I mean something like 15-20% lower performance to make the CPUs stable.

Note that some people want to keep their PCs running reliably for more than 2 years. Or 4.

CPU has never been viewed as a consumable by normal consumers. Or something that by becoming unstable could potentially ruin their data, efforts, plans, cause other damages.

 

[DAPUNISHER]

Yes, but that's water over the dam now......... how does Intel think this is going to play out now?

They can get ahead of the issue and try and keep some customer loyalty or get dragged over the coals which will cost them even more money (and unfortunately any class action lawsuit would benefit the lawyers more than customers)

They knew about the oxidation well over a year ago. That's when they could and should have gotten out in front of it. They showed who they are as a company on that alone. Then they went into full hold my beer mode by throwing their board partners under the bus. Denying the RMAs. Telling DC customers to pound sand. Constantly kicking the can down the road. Allowing GPU and game companies to take the heat for the crashes, bans, and other failures. Providing tech support solutions that require losing the performance you paid for, instead of making RMA your first recourse. To top that off, most of the TS advice has been time consuming, detailed, and confusing for most users.

There will be CA lawsuits no doubt. But it won't stop there. They are going to get hammered by governments almost certainly. They have and will continue to, lose customers, some for good. The platform is officially cursed and if they do release Bartlet for it next year, it's going to be like that scene in Jaws where no one will go in the water first.

Many from tech sites to social media and message board goers, have also shown who they are, by failing to put consumer rights first. And instead linking arms and trying to shield INTC. It's a really bad look, and I for one, won't be forgetting while my axe is still shiny.

 

[DAPUNISHER]

They'll work out deals with their big OEM type customers on the side and the sales reps will take them out to dinner to an extra fancy restaurant that year and then business as usual. Stuff like that has worked for them in the past anyway. (Edit: This very well could come back to bite them, but [hard speculation based on Buildzoid's speculation] it could be how they hope it plays out at this point.)

That would be the final proof for me, that they are in the same trap the old IBM fell in. They don't have the cash reserves or momentum they once did. It is death by a 1000 cuts if they have to keep lowering margins to make sales. They will end up shedding more workforce, the brain drain will continue, and talent will continue to seek happier hunting grounds.

@dmens called all of this fail years ago. I really miss his former employee perspective around here. He was no simp like many that worked there, he was on point and brutally honest.

 

[KompuKare]

Yes, but that's water over the dam now......... how does Intel think this is going to play out now?

They can get ahead of the issue and try and keep some customer loyalty or get dragged over the coals which will cost them even more money (and unfortunately any class action lawsuit would benefit the lawyers more than customers)

Well, Nvidia got off very lightly for bumbgate so that may the model Intel want to follow (as it happen I seem to recall the Nvidia statement "we are mostly a software company" was around the time they really messed up with hardware.)

These CPUs are most important parts of systems, would Intel need to cover the cost of a whole system, when without a CPU it is unusable?

Yes, this had been discussed before.
Say Intel refund someone the full cost of the CPU (or someone takes this up with their credit card company as in many countries they need to help sort this kind of stuff out) where is that consumer then?

Unless they can also return the motherboard, they are either stuck on LGA1700 or moving over to AM5 will incur a large costs.

Which is why the best thing Intel can do is produce a stable LGA1700 part ASAP, or consumers use their refund to buy Alder Lake parts. Yes, some will be miffed by "rewarding" Intel and buying a Alder Lake part but pragmatically it may the easiest thing for consumers.

 

[DAPUNISHER]

Which is why the best thing Intel can do is produce a stable LGA1700 part ASAP, or consumers use their refund to buy Alder Lake parts. Yes, some will be miffed by "rewarding" Intel and buying a Alder Lake part but pragmatically it may the easiest thing for consumers.

Buying 12th gen is throwing good money after bad. No bueno. Intel needs to make it right by those folks without need to spend another penny/pence whatev. And Stockholm Syndrome is a hell of a drug.

Tom and Wendell talked about how OEMs have almost nothing you can change in the bios. That those customers would send in the whole PC for service. I cannot see how Dell and the others would take that on the chin and say thank you sir may I have another.

 

[KompuKare]

Buying 12th gen is throwing good money after bad. No bueno. Intel needs to make it right by those folks without need to spend another penny/pence whatev. And Stockholm Syndrome is a hell of a drug.

Tom and Wendell talked about how OEMs have almost nothing you can change in the bios. That those customers would send in the whole PC for service. I cannot see how Dell and the others would take that on the chin and say thank you sir may I have another.

Ah, even "the best friend money" can buy might not be that friendly to Intel after that!

Yes, it has a bit of Stockholm Syndrome about it but I was talking pragmatically; best of a bad situation for LGA1700 buyers type thing - if Intel want to drag this out and not stand by the products.

Now, the sane thing after being burned is to jump platform but many won't be able without a large financial hit.

Still real Stockholm Syndrome was surely DC customers and Spectre and Meltdown: affected Intel way above all others and the competing solution was not only faster but more energy efficient too. Results: Intel got record DC numbers it was almost like:
"the new patch drops 10% performance what shall we do?"
"Buy 10% more Intel CPUs as nobody ever got fired for buying Big Blue [Intel not IBM these days obviously]."

 

[DAPUNISHER]

I have a feeling that "no one ever gets fired for buying Intel" motto may have hit the wall at speed. Same goes for "Intel just works." Get rekt Jayz2shill.

It isn't a big $ hit if the user is on DDR5 already. Can get 12th gen like performance for gaming with a inexpensive Zen 4 and board. If on DDR4 the 5700X3D and AM4 board can be had dirt cheap and are also close enough to never notice the difference while being about the cost of a 12900K alone here in the U.S. Other markets have to hash out what is the best way to go.