Intel processors crashing Unreal engine games (and others)

BFG10K

Lifer
Aug 14, 2000
22,709
2,971
126
As if we needed any more evidence that these Intel furnace monstrosities are being pushed beyond their limit just to win performance benchmarks. No wonder even 480mm rads or 1.5kg coolers aren't enough for these solar flares.

These parts are being redlined because they're horrifically inferior to AMD X3D processors for gaming, and yet there are still some people on the internet defending Intel's version of Bulldozer.

This is the result of an absolutely incompetent monopoly that sat on 14nm+++ for seven years and was happy to keep selling you +5% performance + new motherboard every 18 months.


 

Dave3000

Golden Member
Jan 10, 2011
1,351
91
91
What are the chances that setting to XMP in the BIOS can play a big part on this instability? Afterall, XMP is considered a form of overclocking as in it runs the memory at speeds faster than what the CPU officially supports, and I wouldn't be surpised if there are people who build their own systems and don't pay attention to what the memory controller on the CPU actually supports and just buy the fastest memory they can afford and that motherboard they buy supports and then set to XMP in the BIOS to get their memory to run at it's advertised speed, even if they don't have the intention of overclocking. Also I used to own a motherboard that if I enabled XMP, it would overclock the all-core turbo speed to it's single-core turbo speed for all core loads.
 
Last edited:

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,542
14,498
136
This is not explicitly Intel fault. Is simply than Motherboards vendors leave unlimited Power Limits or MCE (MultiCore Enhancement) on by default, so you are overclocking out-of-the-box without knowing it.
I don't think most vendors (if any) leave this on by default. Intel experts chime in here ?

Apparantly tomshardware says SOME (but does not say who, or how many) do set this to 4096 watts and 4096 amps, but I find it hard to believe that many would do something so insane.
 

jpiniero

Lifer
Oct 1, 2010
14,586
5,209
136
This is not explicitly Intel fault. Is simply than Motherboards vendors leave unlimited Power Limits or MCE (MultiCore Enhancement) on by default, so you are overclocking out-of-the-box without knowing it.

Unlimited power limits shouldn't cause it to crash. MCE might... but that would mean you got a dud that can't do ACT stably enough on all of the P cores.
 

H433x0n

Senior member
Mar 15, 2023
875
937
96
I don't think most vendors (if any) leave this on by default. Intel experts chime in here ?

Apparantly tomshardware says SOME (but does not say who, or how many) do set this to 4096 watts and 4096 amps, but I find it hard to believe that many would do something so insane.
If you enable XMP you’ve inadvertently enabled MCE for the popular motherboards. This setting means unlimited current and uncapped power limits.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,795
3,621
136
These parts are being redlined because they're horrifically inferior to AMD X3D processors for gaming, and yet there are still some people on the internet defending Intel's version of Bulldozer.
Holy hyperbole!
 

tamz_msc

Diamond Member
Jan 5, 2017
3,795
3,621
136
Most, if not all Intel motherboards have some performance enhancing settings turned on by default.

This has been a problem since the days of Coffee Lake.

Intel just needs to stop allowing concessions to OEMs that are used by them to exploit loopholes in the specifications.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,795
3,621
136
Much ado about numerous vendors shipping their motherboards with bad defaults is not nothing.
Most the people reporting the issue had no idea they were overclocked. Intel can and should put the kibosh on vendors shipping firmware like this.
This has been going on for 7 years at least. It was first observed with 8th gen Coffee Lake. Only it wasn't a problem back then probably because 14nm++ is more tolerant of transients than Intel 7.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,330
4,916
136
I set PL1 and PL2 to 105W on my Z690 rig just for efficiency sake, but I had no idea ICC_max was even a configurable setting.

Once you have to change out of the box settings you're already past 90% of users, whether Intel or AMD...
 

Saylick

Diamond Member
Sep 10, 2012
3,129
6,314
136
I set PL1 and PL2 to 105W on my Z690 rig just for efficiency sake, but I had no idea ICC_max was even a configurable setting.

Once you have to change out of the box settings you're already past 90% of users, whether Intel or AMD...
Yeah, if enthusiasts dont even know about how to fix this issue, average Joe would not stand a chance.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,971
126
As expected, the culprit is ICC_max set to above spec - it should be 307 A for the 14900K.


ICC_max is more important than the PL1/PL2 values.
I'd never heard of this setting until now and had to look it up. I know how to write a double-linked list in C using free() and malloc(). I bet most people in this forum can't do that, so I'm not a computer illiterate moron by any stretch of the imagination.

A lot of tech enthusiasts don't even know about XMP or how to set a high refresh rate,, much less something as obscure as ICC_max.

Much ado about nothing.
Then it should be absolutely no problem for Intel to proactively contact all major CPU reviewers and ask them to retest @ spec. Amirite?
 

tamz_msc

Diamond Member
Jan 5, 2017
3,795
3,621
136
I'd never heard of this setting until now and had to look it up. I know how to write a double-linked list in C using free() and malloc(). I bet most people in this forum can't do that, so I'm not a computer illiterate moron by any stretch of the imagination.
And I absolutely detest writing computer programs (I find it terribly boring - I'd rather do a math problem), and yet I know about ICC_max.
Then it should be absolutely no problem for Intel to proactively contact all major CPU reviewers and ask them to retest @ spec. Amirite?
Yes, obviously.

Then the conclusion of Intel CPUs guzzling power would magically disappear, which will then go against the prevalent narrative.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,971
126
And I absolutely detest writing computer programs (I find it terribly boring - I'd rather do a math problem), and yet I know about ICC_max.
Nobody cares what you find boring because it's irrelevant.

Then the conclusion of Intel CPUs guzzling power would magically disappear, which will then go against the prevalent narrative.
The performance will magically disappear too, that's why Intel happily look the other way when benchmarks are posted.

And it's not really a narrative, more like objective proof. 14900K overall slower than X3D for gaming while absolutely chewing through power. Also throttles @ stock even with a 360mm AIO liquid cooler.

 
Last edited:

coercitiv

Diamond Member
Jan 24, 2014
6,188
11,866
136
I can't feel properly vindicated about this because I know it must have been very painful for people to figure out why these systems were unstable (both users and game devs). Still, it does feel nice to finally have public proof that all this irresponsible board maker behavior is toxic for the consumer. Also, the biggest irony is the folks affected are the ones buying the premium hardware.

Like @zir_blazer said, this is not explicitly Intel's fault. Still, they were complicit. Now they'll pay the PR price.
 

coercitiv

Diamond Member
Jan 24, 2014
6,188
11,866
136
I don't think most vendors (if any) leave this on by default.
Here's how it happens on my MSI Z690 board. Any time I make a change that affects my CPU multiplier or enable XMP, I get this pop-up message in UEFI. You'll notice there is no Cancel button. MSI decided that consumers don't need Cancel buttons, that would only confuse them. And just in case someone tries to say "but", one gets this prompt even if they try to underclock, lower CPU frequency below stock.

power-limit-message.png

So what is Water Cooler setting?! Well, I'm SO GLAD you asked! Let me introduce you to the Power Menu à la carte:

power-profiles.png

Current limits:
  • Boxed Cooler 240A
  • Water Cooler 512A

but I find it hard to believe that many would do something so insane.
I would expect that most boards with Z chipsets and most B/H chipset "gaming" boards with powerful VRM stages will exhibit this behavior. The more expensive the board, the more likely it will behave like this. For the example above, mine was one of the cheaper Z690 models and still does it. In other words, all the boards that one would buy together with a 13900K or 14900K will fit the profile.
 
Last edited:

tamz_msc

Diamond Member
Jan 5, 2017
3,795
3,621
136
Nobody cares what you find boring because it's irrelevant.
Nobody cares if you are a coding wizard either, it's irrelevant as well. But this is a forums for PC hardware enthusiasts, and it reflects poorly on those who flaunt the tag 'enthusiast' if they don't know about the basic things they ought to know about their hobby.
The performance will magically disappear too, that's why Intel happily look the other way when benchmarks are posted.
They perform practically the same if you remove the outlier reviews like Techspot/HWUB (which favour AMD) and some other reviews by various European outlets which favour Intel.

The only difference is the power consumption while gaming - and yes, this thread is about why that is the case.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,971
126
Nobody cares if you are a coding wizard either, it's irrelevant as well. But this is a forums for PC hardware enthusiasts, and it reflects poorly on those who flaunt the tag 'enthusiast' if they don't know about the basic things they ought to know about their hobby.
Nah, it's very much relevant. Someone that builds PCs and codes in C has vastly more knowledge than the average PC user. If you're seriously suggesting all 14900K owners should know ICC_max is intentionally out of spec and causes issues, your argument is pure delusion at its finest.

Your first two posts didn't even mention ICC_max. You obviously furiously ran some google searches after and then posted a follow-up how it's "obviously expected" to be the cause. Oh, you tech searcher, you.

They perform practically the same if you remove the outlier reviews like Techspot/HWUB (which favour AMD) and some other reviews by various European outlets which favour Intel.
"Outliers", because they don't fit the Intel apologist(tm) handbook. Intel could enforce these limits on reviewers and board makers (including mandating PL1/PL2), but they don't. Hmm, what could Intel possibly gain by allowing their hardware to run out of spec for internet benchmarks? o_O

And...I wonder... is it the same idea that was behind their SPEC cheating?


Or how about this one?

 
Last edited:

Thunder 57

Platinum Member
Aug 19, 2007
2,675
3,798
136