Intel processors crashing Unreal engine games (and others)

[igor_kavinski]

The degradation seems to be related to the duration of the stress. Shader compilation and code compilation (especially on Linux) can take anywhere from 20 minutes to hours and these CPUs croak under sustained stress like that.

 

[DAPUNISHER]

The degradation seems to be related to the duration of the stress. Shader compilation and code compilation (especially on Linux) can take anywhere from 20 minutes to hours and these CPUs croak under sustained stress like that.

Time varies greatly between users and model of course. But heavy UE gaming workloads do seem to be how most are exposed to the user.

This poor soul in one of the current threads -

Bought a 13900k. It died after 6 months. RMA got me a brand new one which I sold for 300. Bought a 14900ks. It's currently being RMA'd after a month and a half.
Just gonna leave my 12900k installed for awhile.

I can't give him grief for buying all 3 gens since I do the same thing on AM4. Hell, I change within the same gen on top of it. Only took 6 weeks for the 14th gen to go belly up.

 

[Timur Born]

Which of the UE engine games known to crash do you put a fair number of hours into?

If this were a "UE engine" only problem then I would point to the engine instead of the CPU. But I am always open to put in the work for science, so despite the optics being revolting I will now install The First Descendant.

 

[Timur Born]

We won't know until we understand the nature of failures. We don't know if it's voltage, current, temps, mechanical, factory of origin or a combination of all these factors. Depending on the root cause and possible secondary causes that may accelerate the degradation, a CPU being stable for more than 1 year is no indication that everything will remain fine in the future. (this applies to my 12700K too)

So this is the take? I post a theory ignored by both answers, with the answers being "use UE" and the other "we don't know". So what exactly is the purpose of this thread then?

 

[moinmoin]

If this were a "UE engine" only problem then I would point to the engine instead of the CPU.

It's not a UE or even an engine problem but a data integrity one. One of the tools integrated in UE for data compression (Oodle by RAD Game Tools, bought years ago by Epic for inclusion in UE) does data integrity checks (the horror!) after decompression which is where all those crashes appear to turn up. It seems all other use cases and software just cruise on completely oblivious about potential faulty data caused by this.

 

[DAPUNISHER]

So this is the take? I post a theory ignored by both answers, with the answers being "use UE" and the other "we don't know". So what exactly is the purpose of this thread then?

Your "working theory" goes in the pile with the rest, and it's a big pile. And the reason no one knows for certain what the cause or solution might be, is 100% Intel's fault.

1.The thread serves as a PSA for searchers looking for solutions to their games and apps crashing with raptor CPUs. The crashes often point to drivers, vram limitations, storage, games, etc. But they are all due to the CPU. This isn't in dispute, it's a fact.

2. It provides an unfortunately massive caveat emptor for shoppers considering a raptor build.

3. Intel has acknowledged there are problems and is investigating. It is evidently a complex problem as this has been going on for months.

4. Intel has not been communicating with their customers often or well concerning this, and until if or when they offer up both causes and solutions there is no definitive answers to be had.

 

[DAPUNISHER]

If this were a "UE engine" only problem then I would point to the engine instead of the CPU. But I am always open to put in the work for science, so despite the optics being revolting I will now install The First Descendant.

Hopefully Intel will have answers long before you are done. As to do this correctly, you'll need to put 100s of hours into multiple affected titles over a period of no less than months. And in the end it'll still be a sample size of one. But if it can provide you peace of mind that your CPU is unaffected, it'll be worth it IMO.

 

[KompuKare]

If this were a "UE engine" only problem then I would point to the engine instead of the CPU.

Was going to write something similar to @moinmoin wrote, but they wrote it much better.

But yes, a program which uses checksums. While most just assume the CPU must be correct.

 

[Timur Born]

Your "working theory" goes in the pile with the rest, and it's a big pile. And the reason no one knows for certain what the cause or solution might be, is 100% Intel's fault.

1.The thread serves as a PSA for searchers looking for solutions to their games and apps crashing with raptor CPUs. The crashes often point to drivers, vram limitations, storage, games, etc. But they are all due to the CPU. This isn't in dispute, it's a fact.

Then I misunderstood the intention of the thread. I thought this was a place to discuss the whole mess from all possible angles.

 

[DAPUNISHER]

Was going to write something similar to @moinmoin wrote, but they wrote it much better.

But yes, a program which uses checksums. While most just assume the CPU must be correct.

The fact it doesn't happen on Zen tells you it isn't UE that's nerfed.

 

[Timur Born]

Hopefully Intel will have answers long before you are done. As to do this correctly, you'll need to put 100s of hours into multiple affected titles over a period of no less than months. And in the end it'll still be a sample size of one. But if it can provide you peace of mind that your CPU is unaffected, it'll be worth it IMO.

Hm, I already do have peace of mind for my sample where stability is concerned. Can't read the future for degradation, of course, although 1.5 years of practical use and lots of stress testing does leave the impression of being fine so far.

Concerning the checksum part. What keeps us from running other compression tools like 7-Zip or Rar for hours to check for checksum errors instead of relying on Oodle based gameplay?

 

[Timur Born]

The fact it doesn't happen on Zen tells you it isn't UE that's nerfed.

It tells us that UE/Oodle may use one feature that is broken on Intel CPUs for whatever reason. That's a bit more specific than calling doom on all Intel CPUs for all daily scenarios, with the latter creating unnecessary extra confusion if the former should really be the case.

 

[DAPUNISHER]

It tells us that UE/Oodle may use one feature that is broken on Intel CPUs for whatever reason. That's a bit more specific than calling doom on all Intel CPUs for all daily scenarios, with the latter creating unnecessary extra confusion if the former should really be the case.

This is quickly devolving. I'll make this simple: Many raptor CPUs after a period of time at stock settings become unstable. Intel knows about it. Intel says they are investigating it. This problem that has been ongoing for months has gotten worse instead of better. It does not matter what tasks the CPU can do properly, if they cannot play the games to the level the bigger bar better reviews claim they can, without degrading, they are defective.

 

[Timur Born]

Ok, so you are saying that 1.5 years of up to 16 hours working, gaming and stress testing have no value for my CPU sample, because I ought to specifically use UE. Then you are saying this is a problem of general instability, not specific to UE.

Make up your mind. The person currently devolving this is you, including posting ambivalent laughing emojis at an effort to discuss this in an adult manner.

 

[In2Photos]

Ok, so you are saying that 1.5 years of up to 16 hours working, gaming and stress testing have no value for my CPU sample, because I ought to specifically use UE. Then you are saying this is a problem of general instability, not specific to UE.

Make up your mind. The person currently devolving this is you, including posting ambivalent laughing emojis at an effort to discuss this in an adult manner.

Have you read the thread? Or any of the articles? Watched any of the videos? Just because you don't have a problem with your CPU doesn't mean there isn't a problem.

 

[GTracing]

Ok, so you are saying that 1.5 years of up to 16 hours working, gaming and stress testing have no value for my CPU sample, because I ought to specifically use UE. Then you are saying this is a problem of general instability, not specific to UE.

Make up your mind. The person currently devolving this is you, including posting ambivalent laughing emojis at an effort to discuss this in an adult manner.

I'll say it, your anecdotal evidence has no value to the other readers. Not all CPUs are affected, and even with a bad CPU not all use-cases trigger the issue.

 

[DAPUNISHER]

Ok, so you are saying that 1.5 years of up to 16 hours working, gaming and stress testing have no value for my CPU sample, because I ought to specifically use UE. Then you are saying this is a problem of general instability, not specific to UE.

Make up your mind. The person currently devolving this is you, including posting ambivalent laughing emojis at an effort to discuss this in an adult manner.

You are getting spicy now, please stop. Nothing you type will change anything. Trying to make this about me and you will not change anything.

Facts - There is a problem with Raptor Lake CPUs, and there is emerging evidence that it is a widespread problem that could affect millions of CPUs. Intel is investigating it. Tech press is investigating it. There are a multitude of hypothesis as to what the cause or causes could be. Those are some of the salient points of this thread.

Which I will now be getting back to, as I don't do pointless debating that solves nothing.

 

[coercitiv]

I'll say it, your anecdotal evidence has no value to the other readers. Not all CPUs are affected, and even with a bad CPU not all use-cases trigger the issue.

To understand the scale, Wendel looked closely at 210 data center machines with 13th/14th CPUs in a population of ~2800 machines to which he had access to broader statistics and info from the data centers.

 

[DAPUNISHER]

I'll say it, your anecdotal evidence has no value to the other readers. Not all CPUs are affected, and even with a bad CPU not all use-cases trigger the issue.

Well stated. And that was my sole intent, to provide a workload known to expose the problems.

 

[biostud]

Maybe CPU bins affect which CPUs are more or less affected by degrading and/or instabilities?! I always called my own 13900K "good enough" and in the full range of 13/14th gen it is rather average, right in the middle of VID spread published by Igor's Lab. Maybe that means my CPU doesn't ask for too much (degrading) nor too little voltage (instability) at stock settings and thus stays healthy and stable?!

I have been using my 13900K for over 1.5 years now and send it through various stress tests within mostly sane limits (but know that it hit over 415 A at least a very few times).

If ~50% at some time during there life cycle will become unstable, you do realize that ~50% will not show symptoms? And you could very well be in that group, which is really lucky for you and you should obviously just be really happy that you belong to that group.

 

[Timur Born]

Have you read the thread? Or any of the articles? Watched any of the videos? Just because you don't have a problem with your CPU doesn't mean there isn't a problem.

I have and I know of the various problems. Consider how you came up with the idea that I supposedly said there was no problem, because *I* did not claim that.

 

[Timur Born]

I'll say it, your anecdotal evidence has no value to the other readers. Not all CPUs are affected, and even with a bad CPU not all use-cases trigger the issue.

Indeed, which is why in my very first post I suggested binning based VIDs to be a possible angle to look at. Never ever have I claimed no problem to exists, which goes to show how easily the thread got diverted by other people claiming stuff that was not based on my own post.

 

[Timur Born]

So let's get back to my original post:

"Maybe CPU bins affect which CPUs are more or less affected by degrading and/or instabilities?! I always called my own 13900K "good enough" and in the full range of 13/14th gen it is rather average, right in the middle of VID spread published by Igor's Lab. Maybe that means my CPU doesn't ask for too much (degrading) nor too little voltage (instability) at stock settings and thus stays healthy and stable?!"

Adding to this:
Worse bin = higher VID = degrading?
Better bin = lower VID = instability?

 

[blckgrffn]

Indeed, which is why in my very first post I suggested binning based VIDs to be a possible angle to look at. Never ever have I claimed no problem to exists, which goes to show how easily the thread got diverted by other people claiming stuff that was not based on my own post.

Well, when you post this:

"Ok, so you are saying that 1.5 years of up to 16 hours working, gaming and stress testing have no value for my CPU sample, because I ought to specifically use UE."

We're pretty far in to anecdotal territory and it makes it sound like you're fine and issues must be outside your workflow.

And maybe your speculation is right, its gonna take Intel coming clean with all of their data, of which they probably have quite a lot, to really break it down for us.

 

[GTracing]

Indeed, which is why in my very first post I suggested binning based VIDs to be a possible angle to look at. Never ever have I claimed no problem to exists, which goes to show how easily the thread got diverted by other people claiming stuff that was not based on my own post.

You're not entitled to have other people take your conjecture seriously. Level1Tech's Wendell already addressed the role that voltage plays to some degree. He found that there's more to it than just voltage.