Intel CPU's crashing in games

SpyderTracks

We love you Ukraine
So there's been a lot of upset gathering steam on the inter webs with intermittent games crashes on Intel 13th and 14th gen.

Workarounds range from underclocking to disabling ecores.

BUT, there are a lot of people finding that simply disabling hyperthreading is a proper solution.

The findings so far is that crashes are far more frequent on games developed on Unreal Engine 4 and 5, and I've certainly had a few examples of this where disabling hyper threading immediately corrected the issue.

Affected titles include Hogwarts Legacy, Fortnite, Remnant 2, Nightingale, and more.

No word from Intel as yet, but it does seem there may be a correlation as Intel have confirmed they're removing hyper threading from CPU's from now on





 

ubuysa

The BSOD Doctor
I've been helping a user on another forum with BSODs and freezes whilst gaming. The user is running an i9-13900K and an RTX4080 on an Asus Z790 Gaming A board. We'd managed to highlight the CPU as a likely cause and the user brilliantly came back with the solution. It's apparently a known issue with i9-13900K and i9-14900K CPUs. The solution (or workaround) is from Intel's own forum here...
Please try the following steps:
-Install XTU, run AVX2 test, and check if the processor passed that test, if it fails, please change the next BIOS settings:
Advanced (F7)
AI Tweaker
SVID Behavior: Change to “Intel’s Fail Safe”
-Reboot the OS and run XTU test again in order to confirm if the unit passed the AVX2 test
-Run the games as well and see if the issue happens again
Responses from this thread on the Intel forum are positive and the user I was helping also reports that this SVID change eliminated the BSODs and freezing.
 

SpyderTracks

We love you Ukraine
I've been helping a user on another forum with BSODs and freezes whilst gaming. The user is running an i9-13900K and an RTX4080 on an Asus Z790 Gaming A board. We'd managed to highlight the CPU as a likely cause and the user brilliantly came back with the solution. It's apparently a known issue with i9-13900K and i9-14900K CPUs. The solution (or workaround) is from Intel's own forum here...

Responses from this thread on the Intel forum are positive and the user I was helping also reports that this SVID change eliminated the BSODs and freezing.
Looks like @ubuysa is entirely on to something here.

Turns out returns of Intel 13th and 14th Gen are so frequent in Korea that Intel have warranted opening an investigation into finding out the cause.




For Intel to open an investigation shows that they acknowledge there's a problem, Intel almost never do that, since the 12th Gen with the enormous heat causing warping, they've denied any kind of issue that's affected each platform since, but looks like there's just too much evidence now not to raise an official investigation.

Early reports are that due to the ridiculously high temperatures, the silicon is actually starting to degrade after around 3 months of use literally leading to core meltdown! Nothing has been said officially though so take everything with a pinch of salt at this stage.

If you are facing any kind of instability with your Intel chip and don’t have any luck with @ubuysa thread above, undervolting should be the first step to see if you can find some stability, IMHO, under clocking is totally rediculous and if you're finding that's the only way to get it stable, I'd seek an RMA rather than settling with nerfed performance on a very expensive CPU!
 
Last edited:

Scott

Behold The Ford Mondeo
Moderator
Should just point everyone looking at a 14900k build to watch the last 5 minutes of that vid. It sums it up for me. Amps & Temp pushed way further than the silicone can handle in order to beat AMD at metrics. No consideration whatsoever to the consumer for the £600 paperweight they could end up with.

Added bonus of a CPU that can potentially be held inside 100C with an AIO too.

Curious if Puget are running Intel governed limits on their testing benchmarks etc.
 

SpyderTracks

We love you Ukraine
Should just point everyone looking at a 14900k build to watch the last 5 minutes of that vid. It sums it up for me. Amps & Temp pushed way further than the silicone can handle in order to beat AMD at metrics. No consideration whatsoever to the consumer for the £600 paperweight they could end up with.

Added bonus of a CPU that can potentially be held inside 100C with an AIO too.

Curious if Puget are running Intel governed limits on their testing benchmarks etc.
The other question is so far, this seems to only have been reported in the gaming community, but judging by the findings so far, I think it’s likely that professional users may have stronger cooling setups that are perhaps delaying the degradation, if not preventing it with hardcore coolers.

But I suspect this will become more widespread outside of gaming within 3 months.
 

SpyderTracks

We love you Ukraine
Unofficial update from Intel to board partners made public by Igor of Igors Labs, they blame board manufacturers out of the gate, but admit the root cause hasn't been identified.

Bear in mind a lot of these issues occur with brand new CPU's out of the packaging. The over voltage settings set by some board manufacturers could certainly contribute to deterioration over a period of time, but the issue isn't limited to that.

This will reduce performance slightly putting in proper limits.


The other thing I don't understand about this is that a lot of people have said that disabling hyperthreading has stopped the issue from occuring, but without applying the proper limits on the board power profile. I don't understand enough about the workings of CPU's at this level, but don't really understand the correlation between hyperthreading and ramping up excess voltages. Intel have been pushing to do away with Hyperthreading since 9th Gen, and only haven't due to public backlash, but it's been loosely confirmed that 15th Gen onwards won't have hyperthreading.
 
Last edited:

SpyderTracks

We love you Ukraine
Update from Hardware Unboxed calling out Intel for blaming board partners.

I edited my initial previous post because I’d put a line in the first line saying “great way to annoy your partners” because I suspected Intel had been pushing partners to eek out as much performance as possible so they could remain competitive in reviews.

What’s even weirder is that ASUS and Gigabytes newly released BIOS updates that apply “Intel Baseline Profiles” are different with gigabytes being more aggressive, which is extra weird

But performance wise in multithreaded, you’re looking at anywhere from 9% on Asus to 25%!!! On gigabyte performance drop with the new limits. Gaming is less affected as it’s usually only a few cores peaking in short bursts rather than a sustained all core load, here it’s up to around 5% on Asus and sometimes 12% on gigabyte.

Now currently both the gigabyte and asus new BIOS are beta so don’t expect this to be the final result. We don’t know why gigabytes cuts are so much heavier, weather that’s a mistake or for another reason.

HUB are calling them out saying it’s intels responsibility and that they have evidence to prove it

 
Last edited:

SpyderTracks

We love you Ukraine
Now, this isn’t limited to Intel, AMD had exactly the same issue with the SoC voltage running too high on boards on the new AM5 boards.

The difference in handling it though are entirely different. AMD are very connected to the community and use their forums to track feedback and respond to user issues. They took ownership and rolled out a FEW Agesa updates before they got it right. Board partners in the main were quick to roll out BIOS updates. Asus seemed to struggle more than other partners in getting their boards stable with the new agesa limits and initially they even said anyone applying these BIOS fixes would void their warranty until GamersNexus called them out about it.



Here is also a direct quote from one of the early users affected who worked with AMD and their experience


But with AMD, the release was October reports started hitting in around April, fixes started flowing in April after AMD liaised directly with people affected and people like GN and buildzoid who were investigating, and after a few mis hits, the final agesa update was pushed out to all board partners by beginning of May


With Intel, the officially responded to consumer complaints in February


Yesterday was intels first official response almost 2 months later and that was to try to deny it was anything to do with them and blame their partners which it’s been proven by HUB that they’ve always said these parameters were still considered in spec. And the fixes so far have been done by board manufacturers, not Intel

AMD are not faultless, and this highlights the need for both of them to clearly define what is to be in spec going forwards

But there is a big difference in the attitude of both companies in respect to listening to the community feedback, and their respect for partners.
 
Last edited:

Scott

Behold The Ford Mondeo
Moderator
Really great info, thanks for the summaries as well. Incredibly helpful, especially for anyone without their finger fully on these pulses.

As you have highlighted, everyone makes mistake. It's the reaction that matters more than the issue.
 

Scott

Behold The Ford Mondeo
Moderator
Didn't watch all of it, had a little nosey through it though and caught some snip discussions. Pretty fair from what I could see. They're much more polite with the fanboi contingent than I would be. Imagine criticising an outlet for bringing news of an issue to you purely because you feel the need to justify your choice!
 

SpyderTracks

We love you Ukraine
Didn't watch all of it, had a little nosey through it though and caught some snip discussions. Pretty fair from what I could see. They're much more polite with the fanboi contingent than I would be. Imagine criticising an outlet for bringing news of an issue to you purely because you feel the need to justify your choice!
I got in a twitter argument with a couple of fanboi's who were adamant that as long as intel were offering to fix them, then any voltage values were within spec.

And one of them went on to tell me how it was important that board manufacturers allowed you to adjust the TDP of the CPU????

I called them a few names and exited the conversation.

The very real consideration is that if intel double down on blaming board partners, it squarely forces the responsibility onto them, and Intel can claim that it's not covered by their warranty and people have to go to board partners. That would be suicide for Intel, but I wouldn't put it past them.
 
Top