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.

 

[Storm-Chaser]

then how does it classify as a professional system? Just because a professional use it?
Then if a professional sushi chef uses a 2 dollar knife, is that the same as a professional 2000 dollar Japanese sushi knife?

Its like if a car was designed as family sedan (unmoded)... how is it a Nascar race car?
Same concept.

the 13900f was not designed for professional, and working enterprise enviorments.
They won't even use it in studios, unless its to debug a game or software meant for consumers.
There are Xeons Big and Small, Thread Rippers, and EYPC's for "professional" environments, and they have there own rules, unless you go into HEDT, where some of them are similar, but they are still different from real production machines.

Can a professional use a consumer cpu?
Sure... he can... can he use it for professional settings... i don't think he would like to unless its his only option.
Will businesses use consumer stuff... only on the low low end like cubical office.
Will studio's and production companies use consumer..

NO... there on the big boys, or Apple, which is a hybrid of professional / consumer.

Coming from someone who has owned many xeons and server chips throughout the years, there is honestly not much difference between the two in most cases and this still holds true to this very day. For example, they are both built from the same technology process and share the same basic die, so that's a pretty strong similarity right out of the gate. Generally speaking Intel is just adding cores, cache and a host of additional io stuff to the same die that comprises the consumer chips as well. Server chips typically have superior voltage requirements and may be binned better than their consumer counterparts. But most cannot be overclocked so that's a moot point. I think what you are failing to see is that the 13900K is a very powerful processor and would do just fine and perhaps better in some areas due to higher clocks from the factory.

 

[Storm-Chaser]

You are trying to make everyone think the 13900k is like a disposable camera or a Ford powerstroke 6.0L

 

[aigomorla]

You are trying to make everyone think the 13900k is like a disposable camera or a Ford powerstroke 6.0L

If you want to talk in car sense...

The 13900K is a ecoboost v8 that same kind a F150 uses.
The W-3400 series is a Peterbuilt Cummin's

You can't even put them in the same category because of the immense PCI-E lanes count, and 8 channel DDR5 support on the W series, because it was intended purely for Professionals.

Like you would never see a road car thats not a truck packing a cummin's unless it was to gain views on youtube.

Coming from someone who has owned many xeons and server chips throughout the years, there is honestly not much difference between the two in most cases and this still holds true to this very day.

Cou*ECC RAM SUPPORT*gh...

I don't think i need to say more about Xeon's vs i3-i7's except that one gen when i3's could use ECC.
And if you say ECC is not needed for real professionals, oh man... i think im done with this conversation then.

 

[A///]

If you want to talk in car sense...

The 13900K is a ecoboost v8 that same kind a F150 uses.
The W-3400 series is a Peterbuilt Cummin's

You can't even put them in the same category because of the immense PCI-E lanes count, and 8 channel DDR5 support on the W series, because it was intended purely for Professionals.

Like you would never see a road car thats not a truck packing a cummin's unless it was to gain views on youtube.

I'm having flashbacks to the article i read earlier today on pc gamer by Chris Szewczyk about why m2s suck and u2 should be the way to go. how does this relate to what you two are talking about? it doesn't except u2 ports exist on server boards where they dont on consumer or workstation boards. i hinted at u2 eventually coming to prosumer boards in future due to the space savings they would provide for mobos as of current.no more chunky heavy boards with enough aluminum in them to make a car out of.

the market would go to u2 if if it came in rgb and other garish designs, they'd ignore the ugly cables it has if it means spending less on a board that's lighter and the u2 cases are already large heatsinks. what's the stuff young computer people like? shirtless anime women? Slap on some designs on a u2 case that's mounted vertically and you've got yourself a real winner there.

 

[aigomorla]

I'm having flashbacks to the article i read earlier today on pc gamer by Chris Szewczyk about why m2s suck and u2 should be the way to go. how does this relate to what you two are talking about? it doesn't except u2 ports exist on server boards where they dont on consumer or workstation boards. i hinted at u2 eventually coming to prosumer boards in future due to the space savings they would provide for mobos as of current.no more chunky heavy boards with enough aluminum in them to make a car out of.

Ignoring the fact that most U.2 have about 10x the write endurance a typical M.2 nVME has?
Wait... isnt that enough on its own for a reason why U.2 is more designed for professionals.

 

[gdansk]

I'm having flashbacks to the article i read earlier today on pc gamer by Chris Szewczyk about why m2s suck and u2 should be the way to go. how does this relate to what you two are talking about? it doesn't except u2 ports exist on server boards where they dont on consumer or workstation boards. i hinted at u2 eventually coming to prosumer boards in future due to the space savings they would provide for mobos as of current.no more chunky heavy boards with enough aluminum in them to make a car out of.

Already happened. I have an Asus X570 board with a U.2 connector albeit a single one. and I use a normal M.2 drive (with an adapter).
But aren't EDSFF drives replacing U.2 drives in the server space?

 

[A///]

Ignoring the fact that most U.2 have about 10x the write endurance a typical M.2 nVME has?
Wait... isnt that enough on its own for a reason why U.2 is more designed for professionals.

Does it? I had no idea. what's the bom for either side of the plane? I think it may be the future but what the hell do I know? I thought the internet was a fad at first. if nand costs get lower and lower and u2 becomes a replacement for hdd assuming it exists in the future or adapter breakout cables exist for u2 to sata then it might replace sata ports on boards in the future. I'm sure the tech already exists but I'm betting someone here will look into it and inform us all because I get very lazy with this stuff. such as vvvvvvvv

Already happened. I have an Asus X570 board with a U.2 connector. Albeit a single one and I use a normal M.2 drive (with an adapter).

Really? I didn't know one existed. can you show it to me? idk where you could get a brand new u2 drive now but plenty exist on ebay. used ones

 

[gdansk]

Does it? I had no idea. what's the bom for either side of the plane? I think it may be the future but what the hell do I know? I thought the internet was a fad at first. if nand costs get lower and lower and u2 becomes a replacement for hdd assuming it exists in the future or adapter breakout cables exist for u2 to sata then it might replace sata ports on boards in the future. I'm sure the tech already exists but I'm betting someone here will look into it and inform us all because I get very lazy with this stuff. such as vvvvvvvv

Really? I didn't know one existed. can you show it to me? idk where you could get a brand new u2 drive now but plenty exist on ebay. used ones

It's this board https://www.anandtech.com/show/14657/the-asus-pro-ws-x570ace-motherboard-review/4

 

[A///]

It's this board https://www.anandtech.com/show/14657/the-asus-pro-ws-x570ace-motherboard-review/4

oh very nice. is this a sage board or is that another asus type board? I would love to see more lanes be added in future computer part designs to offer flexibility between m,2 nvme and u.2. 2 nvme tops with the rest on u2 and then satas for those who want big rust spinners.it's a fine balance between doing it for the masses and the take rate by the masses. I dunno how much a brand spanking new u2 drive costs but if the prices of ssd tech came down it would be good to see large storage. relying on nvmes ot increase in size to current large format drives is very much a dream and not likely to ahppen this decade.

 

[gdansk]

oh very nice. is this a sage board or is that another asus type board? I would love to see more lanes be added in future computer part designs to offer flexibility between m,2 nvme and u.2. 2 nvme tops with the rest on u2 and then satas for those who want big rust spinners.it's a fine balance between doing it for the masses and the take rate by the masses. I dunno how much a brand spanking new u2 drive costs but if the prices of ssd tech came down it would be good to see large storage. relying on nvmes ot increase in size to current large format drives is very much a dream and not likely to ahppen this decade.

It's still NVMe compatible. Just a different connector. Or you can break it out into 4 SAS/SATA connections. Works either way. I'm not sure what the future holds - I doubt it is U.2 or U.3 because it's going away in the server space already. But it was a flexible option for that consumer board from that era (2019).

The U.2 drives are expensive because they are for the server market. Highly overprovisioned, high write endurance.

 

[A///]

It's still NVMe compatible. Just a different connector. Or you can break it out into 4 SAS/SATA connections. Works either way. I'm not sure what the future holds - I doubt it is U.2 or U.3 because it's going away in the server space already. But it was a flexible option for that consumer board from that era (2019).

The U.2 drives are expensive because they are for the server market. Highly overprovisioned, high write endurance.

yeah it uses pcie lanes. it would require an evolution on the consumer cpu and board for it to happen, what costs may com down may go back up again due to the nature of more rd costs or fabbing complexity. I have not heard of u3 tbh and if u2 is going away in servers what is replacing it? surely not harder to cool nvme on its own? yes that is their beauty but i suspect the price would come down as consumer side will go up or they would package consumer grade nvme in a u2 package with 60% less endurance and it'll still be a plus over current nvme. there needs to be an evolution from current nvme design and pcie5 is not worth the money at the moe for wha you get but also because directstorage is a feature i do not expect to be fully implemented anytime soon in the next 5 years.

 

[igor_kavinski]

surely not harder to cool nvme on its own?

Posted about Phison E26 controller based SSDs in the Storage section. Lack of cooling leads to file system corruption!

 

[A///]

Posted about Phison E26 controller based SSDs in the Storage section. Lack of cooling leads to file system corruption!

that the faulty controller corsair's been using that got picked up by the tech press 2-3 days ago? normally I'd say go with samsung in any instance but they've been huffing the super glue for a year now.

 

[Storm-Chaser]

If you want to talk in car sense...

The 13900K is a ecoboost v8 that same kind a F150 uses.
The W-3400 series is a Peterbuilt Cummin's

You can't even put them in the same category because of the immense PCI-E lanes count, and 8 channel DDR5 support on the W series, because it was intended purely for Professionals.

Like you would never see a road car thats not a truck packing a cummin's unless it was to gain views on youtube.



Cou*ECC RAM SUPPORT*gh...

I don't think i need to say more about Xeon's vs i3-i7's except that one gen when i3's could use ECC.
And if you say ECC is not needed for real professionals, oh man... i think im done with this conversation then.

Facepalm

 

[Storm-Chaser]

That's the 5.0l coyote engine no such thing as a v8 ecoboost.

You don't deny the fact that both chips share 5he same architecture, do you?

You don't see 140gb per second bandwidth with four channels sufficient for professional use?

 

[Storm-Chaser]

How about a score of over 19000 on cpuz? Not enough power there for professional use?

 

[DAPUNISHER]

@Kocicak called attention to the problem back in 2022. Those pushing back have a plate of crow waiting for them.

 

[DrMrLordX]

@Kocicak called attention to the problem back in 2022. Those pushing back have a plate of crow waiting for them.

Nah but he definitely earned that 3600x2 he always wanted!

 

[DAPUNISHER]

Nah but he definitely earned that 3600x2 he always wanted!

Unlike Captain America, I don't understand the reference. And you can't be saying he was wrong, because that debate is lost before it starts.

 

[coercitiv]

@Kocicak called attention to the problem back in 2022. Those pushing back have a plate of crow waiting for them.

And you can't be saying he was wrong, because that debate is lost before it starts.

He WAS wrong and you would indeed lose the debate before it started.

• he overclocked and literally cooked his CPU with subpar cooling
• never established a baseline for stability
• ordered several CPUs, overclocked them and then returned them to the seller (so good-bye evidence and further testing)

Here's how thermal testing went:

HEY! Put the warning that multithread test runs multiple times in the main instructions and information text! I run it first at unlimited power and prayed that it ends and finally had to shut it down when I smelled things burning!!!! You nearly killed my computer!

His habits related to CPU purchasing and "binning" of Ryzen 5000 series, you should be able to find something similar about 13th gen, AFAIK he bought more than one 13th gen and "settled" for a 13600K in the end:

Despite shelves being empty and reading a ton of frustrated comment at the retailers sites, I managed to order and get ready for pick-up eleven or twelve 5000 series CPUs, all for list price. So it must not be as tragic as it seems on surface.

I picked up 5 of them, returned 4, kept 1 - 5600X. Retailer lured me in picking the first one by sending me the promo code by mistake early. I felt obliged to pick the CPU up. That started my buying, testing and playing spree. BTW I have not used or resold the codes. My moral standards are very, very high.

The whole idea about proving Intel CPUs are unstable relies on doing it at stock settings, by people who don't make a habit of returning dozens of CPUs because they're not golden samples.

What happens the next time some boy cries wolf after overcloking his CPU on day one because reasons?

 

[DrMrLordX]

Unlike Captain America, I don't understand the reference.

Pretty sure he requested AMD make a 12-core Zen2 with two "bad" CCDs called the 3600x2, it's an old joke I dig up every now and then where appropriate. In fact he made a thread for it, though I can't find the thread at the moment.

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...in-the-situation-of-chiplet-scarcity.2570862/ here it is.

It's not really on-topic so I'll leave it at that.

 

[igor_kavinski]

"settled" for a 13600K in the end:

Wonder if it still posts

 

[DAPUNISHER]

I won't comment about the OPs specifics, but one thing is clear to me. Intel over the last few generations are binning closer than ever to try and win benchmarks. They factory overclock to the max. I could certainly see that whatever voltage is required to get 100 mhz over stock could degrade the CPU if done long term, and that could even be hours or days.

the 9600kf and the 1705t were NOT in that category. Any 13900k IS in that category.

This aged extremely well.

 

[DAPUNISHER]

He WAS wrong and you would indeed lose the debate before it started.

• he overclocked and literally cooked his CPU with subpar cooling
• never established a baseline for stability
• ordered several CPUs, overclocked them and then returned them to the seller (so good-bye evidence and further testing)

Here's how thermal testing went:


His habits related to CPU purchasing and "binning" of Ryzen 5000 series, you should be able to find something similar about 13th gen, AFAIK he bought more than one 13th gen and "settled" for a 13600K in the end:



The whole idea about proving Intel CPUs are unstable relies on doing it at stock settings, by people who don't make a habit of returning dozens of CPUs because they're not golden samples.

What happens the next time some boy cries wolf after overcloking his CPU on day one because reasons?

Some of the points you made I concede. The part about stock, I think is a poor choice of words. If you mean without user intervention, that works better. I also don't agree he was wrong.

The weakness in the CPU may be of the kind it will allow short term 100% functionality, but it will manifest itself fully only after slight stress. I do not believe that CPU could have been long term reliable, because the process at lower voltage would be happening too, just slower. I already wrote that what failed at 1.45V and 6000MHz in minutes would have failed anyway at 1.3V and 5500 MHz in tens or hundreads of hours (or even earlier).

One member of overclock.net describes quick degradation too.

Good enough for government work as the expression has it.

 

[coercitiv]

The part about stock, I think is a poor choice of words. If you mean without user intervention, that works better.

I meant stock as in non-overclocked. He pushed the chip to 6Ghz on 3 cores and 6.2Ghz on 1 core, "stock" is 5.8Ghz. Whatever evidence he had was tainted by user intervention, especially because the chip was returned before we could ask for further testing. I did not mean stock as in the setting Intel is only now enforcing. Whatever the board did after pushing the XMP button would have been fine, because it represented the average user experience.

In hindsight it's obvious the stability problems existed back then too, what I'm arguing here is one person can arrive at the correct conclusion based on the wrong evidence, and that person is still considered to be wrong.