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.

 

[Carfax83]

I have my Core i9 13900K set to manual and static all P cores 5.6GHz and VCORE at 1.325V LLC 6/8 on an MSI Z690 Unify X. All e-cores are disabled. Hyper threading is left enabled. Frequency is static and fixed. No power limit set

Under load using CInebench R23, POUT value peaks at 189 watts and VOUT is like 1.257volts under load with a little VDROOP. Idle voltage is like 1.31 per VOUT value in HWInfo64.

Running CPU-Z Stress test, temps peak at 86C and VOUT value like 1.26 under full load and POUT value at 160 to 170 watts.

Am I at serious risk of degrading my 13900K at these settings.

Manual voltages and static clocks aren't harmful provided you have good cooling and don't use excessive voltage, but I would try to lower the voltage as much as possible to reduce the temps.

Raptor Lake is the most responsive CPU I've ever seen for lowering the voltages. You should be able to get a lower voltage than 1.325v. It's not excessive, but you should be in the 1.2xx range.

Mine does 5.2ghz with just 1.066v at load. Load voltages are actually easy to optimize, it's the idle voltages that are the problem from my experience.

 

[Kocicak]

Raptor Lake is the most responsive CPU I've ever seen for lowering the voltages.

Is this good news or bad news? What if the large overvolt out of the box is there to compensate for the quicker than usual degrading?

 

[DrMrLordX]

Am I at serious risk of degrading my 13900K at these settings.

Short answer: unknown. Though what you should really do is return it to stock (with e-cores disabled still), set it to unlimited power, and see what kind of voltage it gives the Raptor Cove cores under those circumstances. Then stay at +10% of that or lower for workloads of that current draw or lower.

If you find something higher in current draw then you need to reassess your calculations.

 

[Wolverine2349]

Manual voltages and static clocks aren't harmful provided you have good cooling and don't use excessive voltage, but I would try to lower the voltage as much as possible to reduce the temps.

Raptor Lake is the most responsive CPU I've ever seen for lowering the voltages. You should be able to get a lower voltage than 1.325v. It's not excessive, but you should be in the 1.2xx range.

Mine does 5.2ghz with just 1.066v at load. Load voltages are actually easy to optimize, it's the idle voltages that are the problem from my experience.

Well I could not even run 5GHZ all core stable at 1.2V and needed 1.25V even LLC6. Maybe 1.225v could have been stable, but at that time was just trying to ensure IMC could handle DDR5 XMP at 6600 so started slow first. Maybe it was ring at 5GHZ 1.2 could not handle.

But anyways, I needed 1.325V to be stable with all. I was close to stable with most with 1.3, but it failed or BSOD Y Cruncher or Prime95 and maybe LinPack XTREME as well ( do not remember), but it did pass OCCT Large Data Set variable and AIDA64 stability test and CInebench at 1.3. But the others it was inconsistent so had to up Voltage to 1.325.

You are able to do 5.2GHz with just 1.066v at load?? What is your ring at? You must have a super well binned chip. Though maybe is your LLC different. My load voltage at 1.2V LLC 6/8 at 5GHz I think was like 1.15 or maybe lower and not quite stable as OCCT Large Data Set Variable froze.

 

[Carfax83]

Is this good news or bad news? What if the large overvolt out of the box is there to compensate for the quicker than usual degrading?

That doesn't make any sense. More volts (along with current) increases the rate of degradation, so if they were trying to reduce degradation the out of the box voltages would be much lower.

But as I've been saying, real degradation usually occurs over a very long time unless you're doing extreme (and I do mean extreme) overclocking/overvolting plus high wattage activities on a regular basis. That will dramatically speed things up.

 

[Carfax83]

Well I could not even run 5GHZ all core stable at 1.2V and needed 1.25V even LLC6. Maybe 1.225v could have been stable, but at that time was just trying to ensure IMC could handle DDR5 XMP at 6600 so started slow first. Maybe it was ring at 5GHZ 1.2 could not handle.

I think you're correct about the ring bus. Mine is at default, and in the UEFI Bios, there is caution about disabling the ring down bus to enable higher ring bus speeds as it may result in overvolting the CPU.

So yeah, I would seriously look into reducing your ring bus speeds so you can lower your voltages. Did you ever run any Aida 64 benches with the ring bus at 5ghz? I'm curious to see how it compares to my scores at default ring bus but with high memory clock speeds.

You are able to do 5.2GHz with just 1.066v at load?? What is your ring at? You must have a super well binned chip.

No, it's just that my ring bus is at default. I saw plenty of other guys on Overclock.net with similar voltages and some that had custom water cooling had high clock speeds as well as voltages that were similar to mine.

 

[Hulk]

That doesn't make any sense. More volts (along with current) increases the rate of degradation, so if they were trying to reduce degradation the out of the box voltages would be much lower.

But as I've been saying, real degradation usually occurs over a very long time unless you're doing extreme (and I do mean extreme) overclocking/overvolting plus high wattage activities on a regular basis. That will dramatically speed things up.

I think he means that by building it a big overvolt, even though this may speed up degradation there is so much extra voltage that most people will never get to the point where they notice it. Kind of like taking a car that burns 1 quart of oil every 5k miles and having to add a quart between 10 oil changes. Make the sump 6 quarts and viola! Now the car is still burning the same amount of oil but you never know it because you make it oil change to oil change without adding oil.

 

[Wolverine2349]

I think you're correct about the ring bus. Mine is at default, and in the UEFI Bios, there is caution about disabling the ring down bus to enable higher ring bus speeds as it may result in overvolting the CPU.

So yeah, I would seriously look into reducing your ring bus speeds so you can lower your voltages. Did you ever run any Aida 64 benches with the ring bus at 5ghz? I'm curious to see how it compares to my scores at default ring bus but with high memory clock speeds.



No, it's just that my ring bus is at default. I saw plenty of other guys on Overclock.net with similar voltages and some that had custom water cooling had high clock speeds as well as voltages that were similar to mine.

I had assumed 5GHz was easy and just set it fixed there and it struggled at 1.2V as OCCT froze like 15 minutes in.

What does your ring run at in default or does it vary?

 

[Hulk]

I had assumed 5GHz was easy and just set it fixed there and it struggled at 1.2V as OCCT froze like 15 minutes in.

What does your ring run at in default or does it vary?

As you can see from my signature I'm at 1.172V as reported by HWinfo during a CB R23 MT run where my P's average 4.94GHz. They move between 4.9 and 5.0 during the run. That is with a -0.075 offset. I'll run at -0.1 offset up to about 190W but won't make 200W under heavy load.
I don't think I got a golden chip either compared to Carfax83. I have never gotten a really good one in 30 years of buying CPU's! They do all work though...

 

[Carfax83]

I had assumed 5GHz was easy and just set it fixed there and it struggled at 1.2V as OCCT froze like 15 minutes in.

What does your ring run at in default or does it vary?

The default clock speed for the ring bus is 4.6ghz on Raptor Lake, and it will downclock along with the core clock depending on CPU activity. So it maxes out at 4.6ghz when the cores are active, and I think the lowest I've ever seen it is at 800mhz.

That's a big step up from Alder Lake which I think maxed out at 3.6ghz for the ring bus when all the cores were active, including the E cores.

That's why I'm interested in seeing your memory and cache latency+bandwidth scores, because I want to see if overclocking the ring bus is even worth it. So do you have any Aida 64 benchmarks with your ring bus at 5ghz?

 

[Thibsie]

I was one of those overclocking the crap out of core2duo. It converted me to Intel from AMD. I did this for years. I was such an avid overclocker, that what got me elected as a mod here. Until 2017... Anyway, yes, I never killed a cpu, and I put good heatsinks on them. I remember those days. (~2006)

My e7200@3.6G ran with a stupid OEM intel hs/f. Never killed it.

 

[Kocicak]

That doesn't make any sense. More volts (along with current) increases the rate of degradation, so if they were trying to reduce degradation the out of the box voltages would be much lower.

No, they are not trying to reduce degradation, they must assure that the CPU works all the intended life of the product.

For example, if they give it 100 mV overvoltage in the beginning and they run it at 1,25 V the CPU will start failing in 4 years. If they give it 150 mV more, the CPU will still work after 6 years, despite increased rate of degradation!

 

[Mopetar]

Why not just make the voltage dynamic over time so that you minimize degradation while always ensuring that the chip has enough voltage so that it will be stable. Best of both worlds.

 

[DrMrLordX]

Why not just make the voltage dynamic over time so that you minimize degradation while always ensuring that the chip has enough voltage so that it will be stable. Best of both worlds.

That's pretty much what happens when you let the stock boost algo do the work.

 

[Storm-Chaser]

Degradation is largely a myth.

 

[Storm-Chaser]

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.

This is like saying you should only run 2lb of boost when you have 50 on tap.

 

[Storm-Chaser]

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.

Impossible to happen in that short time frame. Look for some other cause. Remember the 13900K has a single core boost of 5.8GHz, so the processor is designed to go reliably to that clock speed, if not more.

Really, no such thing as CPU degradation. I've had multiple systems running very high overclocks for a number of years. The number of CPUs I've dealt with, I would have seen it happen.

Modern cpus are easily good for 50 years, even overclocked, assured you have your ducks in a row.

 

[Kocicak]

Impossible to happen in that short time frame. Look for some other cause. Remember the 13900K has a single core boost of 5.8GHz, so the processor is designed to go reliably to that clock speed, if not more.

Only two specific cores are selected to run at that speed, when I overclocked all to 6 GHz, I pushed some of them by 500 MHz, which may be a lot for some weaker core.

Really, no such thing as CPU degradation. I've had multiple systems running very high overclocks for a number of years. The number of CPUs I've dealt with, I would have seen it happen.

Modern cpus are easily good for 50 years, even overclocked, assured you have your ducks in a row.

There are physical processes as electromigration, which cause CPUs to degrade.

Also there is huge difference in which silicone process are the CPUs built on, robust large processes of the past may have little problem with degrading, but modern processes which are slowly approaching physical limits of what can be done can be more prone to degradation.

On top of that, Intel had a lot of trouble to even make their 10nm process to work at all. It may still be in some aspects problematic.

 

[Storm-Chaser]

Only two specific cores are selected to run at that speed, when I overclocked all to 6 GHz, I pushed some of them by 500 MHz, which may be a lot for some weaker core.

There are physical processes as electromigration, which cause CPUs to degrade.

Also there is huge difference in which silicone process are the CPUs built on, robust large processes of the past may have little problem with degrading, but modern processes which are slowly approaching physical limits of what can be done can be more prone to degradation.

On top of that, Intel had a lot of trouble to even make their 10nm process to work.

No doubt intel was slow to get that going but it did not affect quality. Regarding the 5.8 turbo thing, I'm talking PER core. Meaning, each individual core within the CPU has been tested to be reliable at 5.8GHz. That means each core will not degrade at 5.8GHz. We know this because you don't just have one standout, dedicated core that does 5.8GHz, they all need to be capable of this speed since some workloads jump around from core to core.

Also, Intel would never use a reliability margin of just 100MHz for it's flagship processor.

There is no degradation going on here.

 

[Storm-Chaser]

@scineram

You think processors with a 100 MHz OC are at risk of degradation?

 

[DrMrLordX]

Also, Intel would never use a reliability margin of just 100MHz for it's flagship processor.

Yes they would, especially if it's only for one or two cores in very short bursts before the cooling becomes heat-soaked.

 

[Kocicak]

As I experimented with three 13900K already, from this limited sample I concluded that there may be a 200 MHz safety margin. The first and the current were able to pass 1 thread of Cinebench at 6200 MHz, the second one (the one which degraded) managed "only" 6000. BTW I am very careful with the current one.

 

[Storm-Chaser]

Yes they would, especially if it's only for one or two cores in very short bursts before the cooling becomes heat-soaked.

Okay, I will bite. Name a single other intel processor that degrades at 100MHz OC. And when I say reliability margin, I'm talking point of danger not OC ceiling.
CPUs now adays are smart enough to figure out how to throttle themselves if they get too hot or develop a hotspot before damage occurs.

As I experimented with three 13900K already, from this limited sample I concluded that there may be a 200 MHz safety margin. The first and the current were able to pass 1 thread of Cinebench at 6200 MHz, the second one (the one which degraded) managed "only" 6000. BTW I am very careful with the current one.

Those are good clocks. But attributing that reliability problem to degradation is the real problem here. In reality the processors are just having a difficult time holding those clocks. The "feeling out" process is still going on for 13th gen chips and it seems they cant reliably do above 5.9-6.0GHz all core without VERY good cooling (so you are running out of OC ceiling not safety margin). Did you use a chiller for those CPUs to see how they performed?

 

[DrMrLordX]

Okay, I will bite. Name a single other intel processor that degrades at 100MHz OC.

100 MHz over what? Those CPUs aren't guaranteed to run high clocks across all cores in any workload for an unlimited period of time. Your idea of what amounts to a "100 MHz OC" is completely broken.

 

[Storm-Chaser]

100 MHz over what? Those CPUs aren't guaranteed to run high clocks across all cores in any workload for an unlimited period of time. Your idea of what amounts to a "100 MHz OC" is completely broken.

Over single core turbo clock speed. For example, the 13900K has a 5.8GHz single core turbo, and the logic is this (according to OP):
5.8GHz = no degradation
5.9GHz = degradation
=run at 5.5GHz

Intel would never cut the safety margin to only 100MHz over stock. His logic is akin to saying your car is going to destroy itself almost immediately if you drive over 62mph for any extended period of time.

i.e. My 9600KF has been to 5.8GHz multiple times with no detriment running voltage in excess of 1.6. And has a sustained overclock of 5.3GHz @ 1.400v for nearly two years now. I have punished the rig but it still comes back for more.

My phenom II 1705T x6 was run for nearly 12 years at 4.0GHz with a 3.0GHz NB and little hyper 212 cooler. Didn't degrade whatsoever. Stock clocks were 3.5GHz