Intel 13th/14th instability and degradation topic - Lets talk?

[Memmento Mori]

Hi Guys,

 

just kinda thinking if you have any experience with that.

 

I was intending to delid my 14900K but before i made anything i just read about the degradation problem of some of the CPUs. So there is as well a fix for it like to upgrade bios and make some settings which i did and gues what? As im on air cooler the seetign ust causing i always have the same result in cinebench and the CPU is downclocking it self based on the temp... Which means for me makes no sense to delid it now and test the Thermal grizlys heat spreader alonng with 2 separate coolers... Or am i wrong? Missing something? 

 

Here some media i digged in about it:

 

 

 

 

[ENTERPRISE]

I wouldn't delid it just based on the fact the CPU has a high failure rate and once delidded there is no way Intel would honour a replacement.

[Memmento Mori]

Thanks E. I will probably do it as recomended, its becomming interesting...

 

 

[ENTERPRISE]

13 hours ago, Memmento Mori said:

Thanks E. I will probably do it as recomended, its becomming interesting...

 

 

Thanks, I have actually been waiting for GN to release this video as they eluded to some theories based on insider information they received. Will have to sit back and watch this later. 

 

If it were me and I had a CPU from the affected series, I would be bummed out.

[Kaz]

This is a continuation of the Unreal Engine 5 issue that was popping up with Intel processors.  So far the only thing that has been alluded to is that mobo manufactures were running the chips above power spec.  Wendell seems to think there's more to it than that.

 

My 13700k is unstable with C6 and C7 power states enabled.  It hard freezes quite often.  Problem is non-existent when they are disabled in the bios options.  I discovered this after playing with GNS3 virtualization software.  The instability had me reinstalling windows before I realized I'd changed the bios power states.  I had planned to under volt this chip but never got around to it, in my lazy fix, intel power profile has always been enabled. 

 

I'm in this weird limbo where I don't want it to be a problem because I don't want to figure out a replacement.  I'd probably switch to AMD platform, which means changing my motherboard.  There's another part of me that says I'm way better off dealing with this now than 4 years after I've purchased it when the warranty is gone. 

[Memmento Mori]

Well ill wait what they figure out. I was having this problems like games crashing, but thought that my undervolt and clock settings were just unstable and wated to delid it to get a better termal headroom but would be most probably a mistake… well lets see how this develops and if i get somehow my money back i would go to AMD team… would be beneficial with the pcie line spliting for the nvme pcie card  

[ENTERPRISE]

https://www.tomshardware.com/pc-components/cpus/intel-finally-announces-a-solution-for-cpu-crashing-errors-claims-elevated-voltages-are-the-root-cause-fix-coming-by-mid-august

[Slaughtahouse]

45 minutes ago, ENTERPRISE said:

https://www.tomshardware.com/pc-components/cpus/intel-finally-announces-a-solution-for-cpu-crashing-errors-claims-elevated-voltages-are-the-root-cause-fix-coming-by-mid-august

Thanks for sharing. Pulling this from the PR quote:

 

incorrect voltage requests to the processor

 

It would be interesting to understand this point more. The level of deviation (average) under different loads. I presume this means any voltage monitoring from software was completely incorrect? As many enthusiasts seem to have instability issues even after lowering voltages from what I've read online. 

[Sir Beregond]

Last GN video on the matter seemed to suggest there was possible defects in manufacturing too which if true, this wouldn't fix. Will be interested to hear what happens with that.

 

As for the voltage, will be interested to hear by how much they are modifying it. If the parts no longer can boost to previously advertised speeds as a result of the update, that will be problematic.

 

I know @Mr. Fox and @RageSet have been pushing their 13th and 14th gen i9's pretty hard from some Discord talks. Would be interested to hear if either of you have had degradation issues.

[Slaughtahouse]

31 minutes ago, Sir Beregond said:

Last GN video on the matter seemed to suggest there was possible defects in manufacturing too which if true, this wouldn't fix. Will be interested to hear what happens with that.

 

As for the voltage, will be interested to hear by how much they are modifying it. If the parts no longer can boost to previously advertised speeds as a result of the update, that will be problematic.

 

I know @Mr. Fox and @RageSet have been pushing their 13th and 14th gen i9's pretty hard from some Discord talks. Would be interested to hear if either of you have had degradation issues.

 

Intel has unveiled that in 2023, limited production units of 13th gen CPUs were in fact impacted by oxidization issues. Doesn't exclude other factors discussed, alleged by GN's sources but interesting to see how all this pressure has finally made Intel open up. Hope they* set the course right.

 

Spoiler

Quote

 

Intel statement on via oxidation

 

Short answer: We can confirm there was a via Oxidation manufacturing issue (addressed back in 2023) but it is not related to the instability issue.

 

Long answer: We can confirm that the via Oxidation manufacturing issue affected some early Intel Core 13th Gen desktop processors. However, the issue was root caused and addressed with manufacturing improvements and screens in 2023. We have also looked at it from the instability reports on Intel Core 13th Gen desktop processors and the analysis to-date has determined that only a small number of instability reports can be connected to the manufacturing issue.

For the Instability issue, we are delivering a microcode patch which addresses exposure to elevated voltages which is a key element of the Instability issue. We are currently validating the microcode patch to ensure the instability issues for 13th/14th Gen are addressed. - Intel representative via Reddit.

 

 

Edited July 23 by Slaughtahouse

[Memmento Mori]

 

intel meeeeh

[Kaz]

On 23/07/2024 at 07:53, Slaughtahouse said:

Thanks for sharing. Pulling this from the PR quote:

 

incorrect voltage requests to the processor

 

It would be interesting to understand this point more. The level of deviation (average) under different loads. I presume this means any voltage monitoring from software was completely incorrect? As many enthusiasts seem to have instability issues even after lowering voltages from what I've read online. 

Hwinfo64 shows my 13700k Vcore jumping to 1.35.  I believe this is correct.  Other users have reported theirs jumping as high as 1.4.  I looked into it when I first noticed it, but it appeared to be the standard behavior for the chip.

[J7SC_Orion]

...close call for AMD Ryzen 9k ?

 

 

[Slaughtahouse]

18 hours ago, Kaz said:

Hwinfo64 shows my 13700k Vcore jumping to 1.35.  I believe this is correct.  Other users have reported theirs jumping as high as 1.4.  I looked into it when I first noticed it, but it appeared to be the standard behavior for the chip.

I want to believe too but if it's within the microcode or deeper, I'm not so sure what the sensors are reporting and what is being read out by HWiNFO or any software, is correct. 

 

Hopefully when the update arrives mid August, we'll get more insights. If I were you, run some stress tests and log it in HWiNFO (export the csv) before and after the microcode update. Take those .CSVs, import it into a freeware like Google Sheets, and visualize it to see the difference. 

[acoustic]

Using something like ASUS EC VRM VLATCH MAX is going to give you a good idea of what the chip is actually receiving, at least as accurate as you can get without plugging directly into the motherboard with external tools.

 

I've seen, with stock behavior on a 14900K, on Windows boot with HWINFO64 on Auto-Start, that I will have a VLATCH Max of 1.6x at times.

 

I highly recommend setting an IA VR Limit that is just barely above your 2-core boost. This corrected that behavior for me. With manual tuning, I did not see this behavior.

[schuck6566]

About halfway through this vid he shows voltages in a bios utility pulling over 1.6v on the core with Intel's Fail Safe turned on and asus multicore enhancement disabled.

 

[Mr. Fox]

  

On 23/07/2024 at 07:57, Sir Beregond said:

I know @Mr. Fox and @RageSet have been pushing their 13th and 14th gen i9's pretty hard from some Discord talks. Would be interested to hear if either of you have had degradation issues.

 

I have no issues of any kind with my 13th and 14th Gen CPUs. Excluding the use of Atom cores (which breaks AVX functionality) they are the best processors I have ever owned, hands down. I wouldn't want to own anything else as an alternative, Intel or AMD. All three of my LGA-1700 desktops are overclocked 24/7. I think the issues people are having are due to using eTVB (or PBO if you are an AMD fan) and allowing the two preferred cores to boost to the moon with crazy voltage and using adaptive voltage that allows wild swings and spikes. I overclock the way I always have. Sync all cores and set static (manual) voltage at the value needed for stability. My VID and Vcore are close to the same and I run very high LLC so that idle and load have minimal change. The gamerboy approach to overclocking with massive undervolting and manipulating LLC has never been a good approach in my opinion.

 

If I were going to complain about anything with my LGA-1700 CPUs it would be the use of Atom cores instead of just adding more normal performance cores. These CPUs perform wonderfully, but they would be even more awesome with more performance cores with hyperthreading and AVX in lieu of the "efficient" (i.e. weaker) Atom cores that lack hyperthreading and AVX functionality.

 

I do not care for this person. His potty mouth and self-aggrandizing behavior is over the top. But, that said, more often than not I agree with him. If you haven't seen it, watch this video. It applies to Intel and AMD. In my opinion it is the only "right" way to overclock. And, it can make a degraded CPU stable again. The degraded CPU is really no different than a brand new silicon lottery loser that requires more than desirable voltage. No need to worry if you do it this way. This is one of his better "how-to" videos. If you do it from day one (sync all cores and set static voltage) you will likely never see degradation unless you go crazy with high voltage and cook the CPU with insane thermals. Doing that will kill any Intel or AMD CPU, or GPU.

 

Edited July 25 by Mr. Fox

[Memmento Mori]

Thanks for sharing, did the recomended settings and looks good so far. Have to test it a bit more but looks like a good way to as well test the coolers finally and after the deliding… thanks for sharing and if you have any other advice please share.

[Sir Beregond]

On 25/07/2024 at 11:08, Mr. Fox said:

  

 

I have no issues of any kind with my 13th and 14th Gen CPUs. Excluding the use of Atom cores (which breaks AVX functionality) they are the best processors I have ever owned, hands down. I wouldn't want to own anything else as an alternative, Intel or AMD. All three of my LGA-1700 desktops are overclocked 24/7. I think the issues people are having are due to using eTVB (or PBO if you are an AMD fan) and allowing the two preferred cores to boost to the moon with crazy voltage and using adaptive voltage that allows wild swings and spikes. I overclock the way I always have. Sync all cores and set static (manual) voltage at the value needed for stability. My VID and Vcore are close to the same and I run very high LLC so that idle and load have minimal change. The gamerboy approach to overclocking with massive undervolting and manipulating LLC has never been a good approach in my opinion.

 

If I were going to complain about anything with my LGA-1700 CPUs it would be the use of Atom cores instead of just adding more normal performance cores. These CPUs perform wonderfully, but they would be even more awesome with more performance cores with hyperthreading and AVX in lieu of the "efficient" (i.e. weaker) Atom cores that lack hyperthreading and AVX functionality.

 

I do not care for this person. His potty mouth and self-aggrandizing behavior is over the top. But, that said, more often than not I agree with him. If you haven't seen it, watch this video. It applies to Intel and AMD. In my opinion it is the only "right" way to overclock. And, it can make a degraded CPU stable again. The degraded CPU is really no different than a brand new silicon lottery loser that requires more than desirable voltage. No need to worry if you do it this way. This is one of his better "how-to" videos. If you do it from day one (sync all cores and set static voltage) you will likely never see degradation unless you go crazy with high voltage and cook the CPU with insane thermals. Doing that will kill any Intel or AMD CPU, or GPU.

 

Thanks for sharing and I really agree with this approach as well. I do wonder sometimes about my 5900X using PBO, but seems to work fine. Static OC I almost have the same issue as P vs E core where I can easily all core 4.7Ghz on my first 6 cores (1st ccd/ccx), but the 2nd set of 6 cores does better at 4.5GHz (2nd ccd/ccx). The first ccd is definitely much better binned on my cpu. Benching I have gotten 4.8GHz all core to work on the first ccd, but it's not stable there.

 

As for my 13900KS, I really appreciated your advice on Discord and videos. Overclocking felt a lot simpler on say my old 4790k and to say I was overwhelmed with all the options available in my Z790 Dark for the 13900KS would be an understatement.

 

I look forward to playing around with it more.

 

As for the Atom cores, would have definitely been interesting if they did a SKU that ditched those to either be a 10 or 12 Performance core part. I wonder if that was just using too much power and was too hot.

Edited July 27 by Sir Beregond

[RageSet]

On 27/07/2024 at 12:07, Sir Beregond said:

Thanks for sharing and I really agree with this approach as well. I do wonder sometimes about my 5900X using PBO, but seems to work fine. Static OC I almost have the same issue as P vs E core where I can easily all core 4.7Ghz on my first 6 cores (1st ccd/ccx), but the 2nd set of 6 cores does better at 4.5GHz (2nd ccd/ccx). The first ccd is definitely much better binned on my cpu. Benching I have gotten 4.8GHz all core to work on the first ccd, but it's not stable there.

 

As for my 13900KS, I really appreciated your advice on Discord and videos. Overclocking felt a lot simpler on say my old 4790k and to say I was overwhelmed with all the options available in my Z790 Dark for the 13900KS would be an understatement.

 

I look forward to playing around with it more.

 

As for the Atom cores, would have definitely been interesting if they did a SKU that ditched those to either be a 10 or 12 Performance core part. I wonder if that was just using too much power and was too hot.

 

I am pretty much in line with @Mr. Fox While in my reviews I allow my Asus Z690 motherboard to run crazy for an "out of box" experience, for everyday usage or overlocking, I manually set everything. With that said, I wouldn't recommend the 14900K/KS or 13900K/KS for novice PC users or gamers. These CPUs are for tinkerers with time on their hands and a whole lot of cooling. 

[Mr. Fox]

A powerful CPU is not necessary for gaming. The GPU is more important. As long as the CPU is not so weak that it becomes a bottleneck, it really doesn't matter that much. And, if gaming at 4K isn't something that is important, you don't even need a beast GPU. You can have a decent gaming experience at 1080p or 1440p with an inexpensive midrange GPU.

 

Heck, I have been playing games the past couple of days on a Xeon E5-1680 V2 (8C/16T) X79 CPU overclocked to 4.7GHz with DDR3-2400 paired with a first generation Titan Black 6GB GPU and it does fine as long as I run medium to high settings (not Very High or Ultra) and keep the resolution at 1440p or less. No issues at if the game doesn't insist upon having the latest cancer OS and the newest video drivers. Very respectable experience.

Edited July 29 by Mr. Fox

[Sir Beregond]

3 hours ago, Mr. Fox said:

A powerful CPU is not necessary for gaming. The GPU is more important. As long as the CPU is not so weak that it becomes a bottleneck, it really doesn't matter that much. And, if gaming at 4K isn't something that is important, you don't even need a beast GPU. You can have a decent gaming experience at 1080p or 1440p with an inexpensive midrange GPU.

 

Heck, I have been playing games the past couple of days on a Xeon E5-1680 V2 (8C/16T) X79 CPU overclocked to 4.7GHz with DDR3-2400 paired with a first generation Titan Black 6GB GPU and it does fine as long as I run medium to high settings (not Very High or Ultra) and keep the resolution at 1440p or less. No issues at if the game doesn't insist upon having the latest cancer OS and the newest video drivers. Very respectable experience.

Absolutely. I was perfectly happy with my 4790k and GTX 980 until I upgraded off of 1080p.

[J7SC_Orion]

7 minutes ago, Sir Beregond said:

Absolutely. I was perfectly happy with my 4790k and GTX 980 until I upgraded off of 1080p.

 

These days:

 

 

 

Good ol' days 

 

 

[Sir Beregond]

11 hours ago, J7SC_Orion said:

 

These days:

 

 

 

Good ol' days 

 

 

Wow - nice OC with the chiller!

[Kaz]

Thought I'd give an update on the RMA processes for anyone wondering.

Initially intel requested the following information when filling out a RMA request.

• Can you describe the issue more to us?
• Was this working before or a newly integrated system?
• Did you make recent changes to the system?
• Have you performed other troubleshooting steps?
• Was the processor purchased less than 30 days ago?
• Did you overclock the processor?
• What power scheme are you currently running? You can find this option at Control Panel -> Power Options. The power schemes include: Balanced, Power saver, High Performance, or Ultimate Performance.
• To expedite warranty requests, provide the serial number (ATPO) found on the processor markings and shipping box

 

I included all the above information in the initial request, so I didn't respond to their follow up request that requested the same information.  This resulted in them closing the case 5 (3 business) days later.  When I followed up with them by providing the information they reopened their case and directed me to this webpage.

 

June 2024 Guidance regarding Intel Core 13th and 14th Gen K/KF/KS instability reports - Intel Community

COMMUNITY.INTEL.COM

---

Intel and its partners are continuing to investigate user reports regarding instability issues on Intel Core 13th and 14th generation (K/KF/KS)...

 

These are the recommended power specs for anyone running an intel processor.  It is recommended to adjust your motherboard settings appropriately.  Intel has said they will follow up with me in a month to see if those changes make a difference.

 

(Spoiler alert, it's not going to help.)  My motherboard has had Intel stock power limits enforced from the beginning.  The moment I re-enable all my c-state power settings, it's going to freeze within 20 minutes.  The extra time Intel has requested will allow me to see AMD 9000 reviews and decide what to do from there.  I'm curious how the 9900x preforms against the 7900X3D.  I'm very concerned that a 14700K replacement will only add more cores and create a bigger ring bus voltage problem.  I don't want to buy a new AMD motherboard, but I don't want a faulty processor in the next 4 years even more. 

 

Just a side note for anyone else dealing with this, intel replacements do not reset the warranty time.  The last time I went through intel RMA for an SSD, the SSD was replaced and the replacement died 1 1/2 years later.  If given the option between replacement and refund, always choose refund.  You can buy a replacement and get a new warranty with it.