Intel processors crashing Unreal engine games (and others)

[blckgrffn]

Me on that hill

View attachment 105896

If it isn't stable out of box, it's broken. You should be able to throw and go. XMP and EXPO are technically overclocking so changing those is really YMMV with no guarantees.

I guess I'll keep repeating this since the "No one stability tests anymore" pesists. Users including one of the hosts of the full nerd podcast had their degraded CPU pass all the normal stress tests. Ergo, the usual stability testing was unilluminating. Oodle and other compression-decomprssion tasks that are sufficiently rigorous revealed the damage. Jumping Jeebus on a pogo stick, they were telling people to install Nvidia drivers 10 times in a row to test for degradation. 🤣

Love it.

With regards to XMP/EXPO how are mortals supposed to set their memory to the speed advertised on the memory stick box if not for using those settings. There are so many settings - and sub settings - that are embedded in the stick info that thinking people can do it by hand is silly.

XMP/EXPO shouldn't be overclocking at all, that's should be just setting your ram to its advertised clocks/latencies.

I mean, I get it. Doomguy doesn't need XMP/EXPO, he'll just rip and tear until its finished. But I do! I remember the past and so many blue screens and hard BIOS resets as I set the timings manually on fancy DDR. I am over it. Just work, darn it.

Looking back for the quotes was nice trip down memory lane. Maybe I remembered wrong, but back in 2020/2021 we were talking about disabling MCE, etc. Trying to get the boards to play nice with settings we actually set instead of having "optimizations" enabled on one screen that totally overrode what you might put in on another screen. This train wreck has been coming for quite awhile.

Maybe we should be lauding Intel's chips for being so darn tough they were hard to kill in the past.

 

[AdamK47]

He blames the game at one point. I think he is trying to avoid acknowledging the CPU is damaged because it is going to hit him in the pocketbook. GN reviewed a PC from his S.I. biz -

He may have an SI business. Doesn't mean he's all that knowledgeable.

Judging by his video, I'm going to lump him into the vast group of people ignorant to the whole thing.

 

[DAPUNISHER]

He may have an SI business. Doesn't mean he's all that knowledgeable.

Judging by his video, I'm going to lump him into the vast group of people ignorant to the whole thing.

Agreed on his lack of tech acumen.

Note: His chat was telling him what is wrong over and over; that it's raptor lake causing the crashing. It's going to cost his biz money if it hasn't already. I don't know that his biz is big enough for contra revenue or other incentives and credits, but I would not be surprised if they end up eating the costs for RMA'd PCs. His behavior is sus on that basis alone. Besides, his motto is ABC = always be cheating.

XMP/EXPO shouldn't be overclocking at all, that's should be just setting your ram to its advertised clocks/latencies.

Ideally the board would do it automagically. I have long maintained that OEM/ODM rarely have it enabled on Prebuilts because it cuts down significantly on tech support tickets at the modest JEDEC settings.

I have had 2 different DDR4 kits I bought that had manufacturing dates during the pandemic, have a stick go bad in 2yrs or so of running XMP. Patriot Viper 3600 and OLOy 3600. I have had DDR3 that ran overclocked past max spec for 5yrs+ issue free until I sold it. IMO manufacturing issues and poor QA during that period are why hardware is so sketch.

 

[Ranulf]

Re Asmongold and his SI business, LTT's review was the best of the bunch just for the uh, logo criticism. Starforge may need some blue pills for the blue team cpus in their wang computers.

Spoiler: NSFW LANGUAGE

 

[DrMrLordX]

Maybe I'm just too old school, but back in the day when I was getting game crashed due to overclocking or some other BIOS setting, I would run stability test software and tweak settings to get to a point of high confidence in system stability.

What I see now is people running out of the box motherboard overclock settings (MCE XMP and the like) and being completely oblivious to stability problems that exited right away.

Sure, maybe, but it's still supposed to work. Yeah I strongly suspect that a lot of these CPUs would have failed some stability test or another and should have been RMAd. Though if the existing stories about Raptor Lake gaming servers from Wendel's vid (for example) are any indicator of what people are experiencing, it's clear that only certain specific workloads can cause problems and/or that the problems that occur are arcane (such as I/O-related issues). Running normal stability tests might not have exposed this problem. In fact, there are probably enough people that are "old school" such as yourself that DID run stability tests on their 13900k or 14900k with MCE and XMP enabled and found exactly no problems at all, until of course they did find problems running some UE5 game or another (oops).

 

[DAPUNISHER]

Appears to be so many RMAs they have run out of the 13900K and 14900K in some places. Offered a refund so the owner took it and is recommending to other to RMA now to take advantage of the same offer. Not sure if this is in the U.S. or not.

Top comment is gold - "Because we don’t have the i9-13900k, we’re going to give you the i9-14900k! Except we don’t have that either…."

https://www.reddit.com/r/pcmasterrace/comments/1ey0s2u/intel_13900k_refund

 

[Steltek]

Appears to be so many RMAs they have run out of the 13900K and 14900K in some places. Offered a refund so the owner took it and is recommending to other to RMA now to take advantage of the same offer. Not sure if this is in the U.S. or not.

Top comment is gold - "Because we don’t have the i9-13900k, we’re going to give you the i9-14900k! Except we don’t have that either…."

https://www.reddit.com/r/pcmasterrace/comments/1ey0s2u/intel_13900k_refund

"And, because, we don't have any money in stock either, we are going to offer you in exchange this cheap made-in-China fuzzy Intel pillow featuring an image of the elusive, non-defective i9-14900k on one side and Pat Gelsinger's image on the arse, err, back side."

 

[alcoholbob]

That's interesting I remember years ago discovering my 9900K undervolt was stable running everything including all the major benchmarking tools except shader compilation for Unreal games would crash and dialing my undervolt back would be the fix.

Since Asmon's PC is crashing I'm guessing both A) his CPU has degraded and B) he hasn't upgraded to the latest microcode. The first thing I noticed with 0x129 was stock voltages went up like 60mV, which I assume is Intel's "fix" for degraded but not dead Raptor Lake CPUs, which is just pump more voltage to get them stable again, but throw in a 1.55V cap so the degradation doesn't get substantially worse through the extended warranty period from voltage spikes.

 

[Sgraffite]

This would be an ideal time for a shader compilation stress test to come out that isn't part of a game

 

[DrMrLordX]

This would be an ideal time for a shader compilation stress test to come out that isn't part of a game

You know, that's a really good idea. That's no small ask, though perhaps something could be hacked together with UE5?

 

[igor_kavinski]

Street Fighter 5 demo has a pretty long shader compilation stage when it starts up. Haven't timed it but it was long enough that it annoyed me. But I think it needs to be something that is shader compilation PLUS data integrity verification to throw out an error.

 

[DAPUNISHER]

Redditors defending raptor lake be like -

 

[Sgraffite]

You know, that's a really good idea. That's no small ask, though perhaps something could be hacked together with UE5?

Not sure the level of effort on it. It might not be that difficult for someone familiar with UE5 to make a project that doesn't precompile shaders and when it starts will loop on shader compilation during run time. Would that be enough to bring the problem to the surface? I'm not sure if the complexity of the shaders matter or if repeatedly compiling the same simple one(s) over and over would be enough to fish out instability.

Another good test might be repeated Oodle (or equivalent) decompression.

 

[Thibsie]

Street Fighter 5 demo has a pretty long shader compilation stage when it starts up. Haven't timed it but it was long enough that it annoyed me. But I think it needs to be something that is shader compilation PLUS data integrity verification to throw out an error.

Horizon Zero Down was also very annoying, compiling shaders, till they fixed it in a later patch. Almost sorry they did lol.

 

[yottabit]

Not sure the level of effort on it. It might not be that difficult for someone familiar with UE5 to make a project that doesn't precompile shaders and when it starts will loop on shader compilation during run time. Would that be enough to bring the problem to the surface? I'm not sure if the complexity of the shaders matter or if repeatedly compiling the same simple one(s) over and over would be enough to fish out instability.

Another good test might be repeated Oodle (or equivalent) decompression.

This is a pretty funny idea, an "Oops all shaders" game demo

It would be pretty easy to implement in UE5 (at least if you don't care about the package size being large) you could just combine in a bunch of free sample projects and marketplace content to match or exceed the number of shaders in a retail game.

The only thing I'm haven't done before is the compiling shaders at start-up runtime, it looks like Unreal calls this manual PSO Cache bundling. https://dev.epicgames.com/documenta...ches-in-unreal-engine?application_version=5.4

This is the "old way" really, newer trend is towards automated just-in-time shader compilation. This caused stutter in early versions of UE5 but that's been largely corrected I think. Still, bad taste in the devs mouth I'm sure so they bias toward this startup compilation.

Anyways making a compact project (< 1 GB) might be more challenging. I wonder if you could just flush the PSO Cache and regenerate the same few hundred shaders over and over.

It's pretty funny that shader compilation stutter backlash from gamers led to the shift to this start-up "compile all the shaders" and now the gamers get to say "Not like this!"

 

[DAPUNISHER]

It's pretty funny that shader compilation stutter backlash from gamers led to the shift to this start-up "compile all the shaders" and now the gamers get to say "Not like this!"

Only the ones with broken CPUs. Perhaps gamers with potato CPUs as well.

I skimmed through the script (I don't give him views or engagement) for Jay's latest vid on how to tell if raptor lake is degraded. Nothing worthwhile, mostly normal stress testing. I laughed when I read the - Hurr durr! all CPUs degrade herp derp! Yes J, but it is not suppose to happen faster than you fire down a sack of In-N-Out. Thanks for padding the minutes with so much fluff though, very cool.

 

[Sgraffite]

This is a pretty funny idea, an "Oops all shaders" game demo

It would be pretty easy to implement in UE5 (at least if you don't care about the package size being large) you could just combine in a bunch of free sample projects and marketplace content to match or exceed the number of shaders in a retail game.

The only thing I'm haven't done before is the compiling shaders at start-up runtime, it looks like Unreal calls this manual PSO Cache bundling. https://dev.epicgames.com/documenta...ches-in-unreal-engine?application_version=5.4

This is the "old way" really, newer trend is towards automated just-in-time shader compilation. This caused stutter in early versions of UE5 but that's been largely corrected I think. Still, bad taste in the devs mouth I'm sure so they bias toward this startup compilation.

Anyways making a compact project (< 1 GB) might be more challenging. I wonder if you could just flush the PSO Cache and regenerate the same few hundred shaders over and over.

It's pretty funny that shader compilation stutter backlash from gamers led to the shift to this start-up "compile all the shaders" and now the gamers get to say "Not like this!"

It wouldn't even need to be a playable game, you could just compile the shaders and bonus if there is a progress bar and number of passes shown or something. I would have thought it would be fairly compact as you would just need various shader code attached to various materials? Admittedly I have only done a shader one time and that was in Godot.

 

[AdamK47]

He may have an SI business. Doesn't mean he's all that knowledgeable.

Judging by his video, I'm going to lump him into the vast group of people ignorant to the whole thing.

Timestamped:

 

[DAPUNISHER]

It's good to see Starforge has extended the warranty. Though that should be expected with how big their markup is.

 

[DAPUNISHER]

Getting rid of free coffee and snacks seems like such a small thing, but a pebble can start an avalanche. Quote from the comments

https://www.reddit.com/r/pcmasterrace/comments/1f44y3a/intel_moment

"Having worked at Intel for a few years rhe coffee bars were a big thing so was the fruit. Being able to walk 2 min from my desk to a huge coffee bar with 10 different brews and fresh fruit and other drinks was amazing. Always helped me refocus and get back into whatever I was working on. RIP cold brew and bananas."

"Remove coffee from the office where I work and you'll have a mutiny on your hands. ( I'm in the it department of a construction company)"

"Still waiting on my replacement CPUs.

Intel allegedly has no stock on anything.

My 14900KS was f'ed out of the box, started RMA on July 16th, it was approved July 22nd, and only in mid August was I told the 14900KS would be shipping 5-6 weeks, which was updated to "When it becomes available" yesterday.

13900K they discontinued (But left the 12900K, weird eh), they ran out of stock and have quoted the same as the 14900KS to get a 14900K to replace it.

As of now it looks like Intel 15th gen will launch before their shipping estimation, which means Intel is probably just going to try and offer an Arrow Lake upgrade and hope people don't get angry that their motherboards are worthless.

Can't even get a 12900KS because they axed that SKU in July once they realized that anything over a 12900K was f'd.

If I take their refund, that gives me $2050

Then I still have a Z790 Taichi and Z790 Dark Hero that are useless, around $1300 CAD there.

And if I don't buy rebuy Intel, the high binned DDR5 kits I have were just a waste of money. Another $1200, that 96GB 7200 CL34 in my flair is a 4 DIMM 24GB kit, not a lot of those out there so it cost $ and my 8200 CL38 kits are SK Hynix 5600B 3GB M Die with Class A ICs, don't need that on AMD.

This hardware is all under 3 months old now.

F'ing THRILLED.

Their support forums fill up full of complaints that customer service is ghosting them, or something else has gone wrong. pcmr seems to get a thread a day of people in the same boat of waiting, or being offered a refund as their only option."

ROFLWAFFLES!

 

[DAPUNISHER]

The Hitler in the bunker meme I referenced before -

Intel support still has the high case volume banner. I wonder if they have to be kissing so much ass and making so many promises of kickbacks and credits with enterprise clients right now trying to keep from losing contracts. If so margins will not be improving anytime soon.

 

[blckgrffn]

Let's be clear, no one really wanted it to come to this (mass suffering of end users sold defective product) even if we briefly relish in Intel, a big bad for profit corpo, getting the leg sweep.

Remember the alternatives are all big, bad, for profit corpos that might stagnate if they don't have competition.

Seriously though, gutting the coffee and fruit? LMAO.

How about all those stonk buybacks, huh? JUST SHOVELLING MONEY INTO A HOLE.

 

[gdansk]

Let's be clear, no one really wanted it to come to this (mass suffering of end users sold defective product) even if we briefly relish in Intel, a big bad for profit corpo, getting the leg sweep.

Remember the alternatives are all big, bad, for profit corpos that might stagnate if they don't have competition.

Seriously though, gutting the coffee and fruit? LMAO.

How about all those stonk buybacks, huh? JUST SHOVELLING MONEY INTO A HOLE.

AMD survived spinning off its fabs. AMD survived Bulldozer. AMD survived selling and leasing back their own headquarters. AMD survived having a market cap of $2B.

Intel isn't at that level. They'll be fine even if they are struggling.

 

[igor_kavinski]

Intel isn't at that level. They'll be fine even if they are struggling.

But the fruits! The coffee!

Do you know what kind of actions these are?

These are the actions of worthless management refusing to take responsibility for their bad decisions and instead, punishing the worker bees who were only doing what they were told anyway. Now they are making the situation worse by pissing off their workforce who will probably now resort to doing only the bare minimum and not stress it because no one knows who's gonna get fired so why waste time and energy stressing themselves out? Just check in, bide your time, do what's required and get out. No more enthusiasm to show off their talent or good thinking skills.

I'm currently in that mode too coz of what management did to me over the years. I've been picking a lot of fights lately with whoever I think is favored unjustifiably by management, in the hope that they will finally get fed up and fire me so I can move on to doing something better with my life.

 

[adamge]

I returned to this thread and saw a lack of activity. "Is this whole thing a nothingburger," I thought?

But the fruit exasperation is real. Something is afoot here. I'll keep an eye on it.