Can you even properly stress test these CPUs with p95 and all AVX enabled? What cooler would be needed for such task so they don't thermally throttle?

I would assume that some of these errors are from overclocked memory too. A fair bit of people including Intel advertising slides show setups with 8000 memory or higher, but multiple reviewers fail to get 8000 working on even highend mainboards and memory.

People just like to lie to themselves about their greedy OC and think they can achieve stability with low temperatures + extreme voltage. Then Fortnite D3D12 might crash only once in 50 hours, still can't be declared stable...

I do not believe this. I've been playing nothing but UE4 and 5 games this past year and a half. All crashes due to memory OC to 7800mhz and VGA OC. After I got stable, I've been running my 13700kf,13900k for over a year no issues. As for what cooler you need, I have a 240mm AIO for 13700kf at 5.6ghz all core 45ghz e core no throttling. It's called a little luck and some tuning and you don't have to pull 350watts to play games. 60-180watts max is what I have seen on a 13900k at 58,58,58,57,57,56,55. And my CPU sample is not good at all. it's a $400 eBay 13900k but its de lidded to deal with it.

Last time i had a game crash for me was that one game i forgot the name of, its a multiplayer free to play game from former battlefield developers. The game wouldn't run, so I needed to lower the cpu ratio to lower value and then it ran without a problem, shame the game is total trashfest.

I own a 13900K and the only UE5 games I've played were The Finals & Fortnite, neither crashed.

Typhon Six Six Six:

I do not believe this. I've been playing nothing but UE4 and 5 games this past year and a half. All crashes due to memory OC to 7800mhz and VGA OC. After I got stable, I've been running my 13700kf,13900k for over a year no issues. As for what cooler you need, I have a 240mm AIO for 13700kf at 5.6ghz all core 45ghz e core no throttling. It's called a little luck and some tuning and you don't have to pull 350watts to play games. 60-180watts max is what I have seen on a 13900k at 58,58,58,57,57,56,55. And my CPU sample is not good at all. it's a $400 eBay 13900k but its de lidded to deal with it.

intel sold million chips.. it's not only you does have it.. you know.. also, in case you forgot reading and just be mad that this is just a conspiracy to make intel looks bad.. "Collaborative investigations by Epic Games, Nvidia, and various game studios have pinpointed the high-performance capabilities of Intel's processors as the primary cause of these crashes, often resulting in a "Video Memory Exhausted" error. " it's you vs them, then.. i'm grabbing a popcorn

my money would be on bad xmp profiles/related motherboard settings.

This has nothing to do with memory overclock. This issue has been discussed on baidu and chiphell for quite a long time, and I've seen it myself personally (at 5.8 ghz on my sample, as this is past the Stockfish testing limit without risking vcore CPU degradation or staying under 100C; 5.7 ghz is the absolute max here, we'll see how 14900KS does). It has to do with unstable CPU. The shader cache compilations put a rather heavy load on the CPU. It's not a Stockfish AVX2 load (those are brutally hard) but it is more of a load than the transient friendly Cinebench R23 runs, which is actually easier to pass than the extremely old R15, which doesn't even use AVX instructions! Basically, if you can pass Cinebench R15 for 10 consecutive loops (without stopping except to press "run" again), or Cinebench R15 Extreme (a mod which increases the rendering size so it takes about as long as R20 to pass) for about 4 runs, without a crash, appfreeze or WHEA error appearing at the bottom of HWinfo sensors window (or windows event viewer), you should have no crashes in any Unreal engine games. However if you fail R15 but pass R23, you're playing with fire. Note: Stockfish requires as much as 60mv (depending on dev build date and version) or more to pass than R15 does (looping) so keep that in mind. This is for overclocked CPU's but we're dealing with "stock" here--and I use "stock" loosely, as the same thing happened with 13900K during prerelease, with "5.5 ghz" versus "5.4 ghz". Now, "stock" is supposed to be 5.6 ghz on all cores on a 14900K, but due to prerelease ES testing, this was originally, even given from leaked Intel documents themselves, supposed to be 5.7 ghz, but was later downgraded before final release. But all of the ODM's set it at 5.7 on all P cores as this was what the qualification testing was originally using. But some CPUs don't have high enough V/F points on these stock vcore settings to pass at 5.7 ghz, so they end up crashing. Stronger silicon has no problem with this, but weaker chips may fail. Setting the frequency 100 mhz lower on these chips (to 5.6 ghz rather than 5.7) usually passes everything.

EzyPzyJapaneze:

Last time i had a game crash for me was that one game i forgot the name of, its a multiplayer free to play game from former battlefield developers. The game wouldn't run, so I needed to lower the cpu ratio to lower value and then it ran without a problem, shame the game is total trashfest.

The finals? I am having lots of fun with that game.

toyo:

Can you even properly stress test these CPUs with p95 and all AVX enabled? What cooler would be needed for such task so they don't thermally throttle?

Even under custom water loops it's not a challenge getting a 13900K or 14900K to quickly hit 100C. So what water cooler? One that constantly has its water chilled by something. You'd think with how badly AMD have been curb stomping Intel in the custom market they'd throw a bone and use the same socket for more than 3 minutes, but no, they refuse.

They should expand on this, I "thought" i had this problem. Looked all over many forums and reddit, (dx12 out of memory error). Seems to be only people with 4090's and the 13th/14th gens. Luckly I finally fixed my issue with it being my riser cable being a bit to taut and once i gave it alittle more play and reseated my 4090 i never saw the problem again. However it was weird, that if i clocked my 14900k to 5.4ghz all was well, before i found my actual fix. Just a super weird issue. Anyways I think this only plagues people with 4090's and everyone else it seems to be a non issues. I could be wrong tho. Also to add I was getting this on other games like BF2142 when it was happening. Also just for reference Im running my P cores a 6.2x2, 5.9x4,5.7x8, running 8200mhz ram and a mild overclock on my 4090. No issues with crashes at all, Tho i dont play any DX12 UE games.

No issues with my 14900K and 13700K. MCE disabled, Intel limits enforced, and undervolted. I think the issue stems from many motherboards with over the top out-of-box UEFI BIOS default settings.

This is a normal CPU degradation due to the power and temp. Heavy gamers should just RMA every 2-3 months.

joshk:

This is a normal CPU degradation due to the power and temp. Heavy gamers should just RMA every 2-3 months.

Intel will cross ship hardware if you give them credit card information, if not then you will have to wait for turnaround time, either way they will test the hardware and if it is found to 'work' you will be charged or have your hardware returned.

As a long term under-volter and always running below power limits, zero issue here nor any anticipated.

user1:

my money would be on bad xmp profiles/related motherboard settings.

This is basically what it is. Users experiencing instability on amd systems also mostly related to memory/expo settings. I've been there myself as i like to tweak.

alanm:

As a long term under-volter and always running below power limits, zero issue here nor any anticipated.

First time I undervolted was with the 5800x ....if anything the insanity of msi's pbo at 1.375 should not be allowed ever not at the stock settings at least!

New cpus are all space heaters, I keep my 12700k stock and still the cooler will go crazy at times in games that use all 8+4 cores.

alanm:

As a long term under-volter and always running below power limits, zero issue here nor any anticipated.

Venix:

First time I undervolted was with the 5800x ....if anything the insanity of msi's pbo at 1.375 should not be allowed ever not at the stock settings at least!

I think that`s the main problem, the stock settings of some Mbs are simply insane! My Asus Strix B650E-F gaming has a stock voltage of around 1.4v! This would cause the CPU to overheat and not reach the maximum boost speed!!!... Now i`m also undervolting, currently trying 1.15V, and the CPU reachs the advertised 5.4Ghz speed and runs 25 degrees cooler!... AMD and Intel need to stop MB makers from applying crazy voltages to win some useless benchmarks...