News Unreal Engine supervisor blasts 50% failure rate with Intel chips — company switching to Ryzen 9 9900X, praises AMD's praises AMD's single-threade...

[Admin]

Dylan Browne, an Unreal Engine Supervisor and Feature Film VFX at the ModelFarm visual effects studio, posted on X that his company is experiencing a 50% failure rate for systems powered by Intel's Core i9-13900K and 14900K computers.

Unreal Engine supervisor blasts 50% failure rate with Intel chips — company switching to Ryzen 9 9900X, praises AMD's praises AMD's single-threade... : Read more

 

[ThomasKinsley]

Unfortunately the questions I have are the kind of questions that Intel does not want to answer. Are locked chips affected? Do they need the microcode update? How quick does degradation occur on those chips? When chips are breaking and the company is not answering questions, that does not induce confidence.

 

[Alvar "Miles" Udell]

Hmm, 2 weeks ago he posted this on his LinkedIn page:

In the process of planning a fun upcoming project but with the complexity / fidelity I want to be able to produce for it I’d really like to use something like the Threadripper Pro 7995WX (96 core / 192 thread) CPU, I fear its forever out of my reach though!

And now suddenly they have the ability to replace two Intel CPUs which "immediately exhibited instability" as well as "a few others" with unreleased AMD 9950Xs? Something's a bit sus...

 

[hotaru251]

NGL with intel's issues making owners pissed & the vocal statements by big companies about higher failure rate than intel claims....this is likely going to make AMD's market share rise even more.

I do wonder about server side..will people stop buying XEON just because they might of had an issue w/ a core series 13/14th gen? I could see this issue beign domino effect

 

[-Fran-]

Allow me to link this, please:

News - Intel finally announces a solution for CPU crashing errors — claims elevated voltages are the root cause; fix coming by mid-August

Intel announced that elevated voltages are the culprit behind the crashing errors with its overclockable processors. Intel finally announces a solution for CPU crashing errors — claims elevated voltages are the root cause; fix coming by mid-August : Read more

forums.tomshardware.com

As this will just be a continuation of that thread, I'm sure

Regards.

 

[hotaru251]

Hmm, 2 weeks ago he posted this on his LinkedIn page:



And now suddenly they have the ability to replace two Intel CPUs which "immediately exhibited instability" as well as "a few others" with unreleased AMD 9950Xs? Something's a bit sus...

if the project is a personal one then makes sense...he might not want to spend that $ on a cpu.

however work machiens are a company expense & thats where people spend when otherwise might not of.

 

[-Fran-]

Hmm, 2 weeks ago he posted this on his LinkedIn page:



And now suddenly they have the ability to replace two Intel CPUs which "immediately exhibited instability" as well as "a few others" with unreleased AMD 9950Xs? Something's a bit sus...

Vendors have had access to the new AMD CPUs for validation for a good while. That is standard practice, so I'm sure the context here is: "my chosen vendor showed me the performance of the new AMD build they'll get out shortly and they're impressive". At least, that's my initial reading.

Regards.

 

[vanadiel007]

Anything related to voltage supply being to high and degradation, is going to be interesting as Intel might have to replace every CPU out there because there's no way of knowing how bad the degradation is of your CPU until it causes crashes.
And even then you can bet OEM's are going to blame it on everything else besides the CPU because OEM's are responsible for RMA's on OEM CPU's, not Intel.

 

[parkerthon]

The silence is deafening indeed. I sense there’s going to be a huge lawsuits here. obviously this issue hasn’t been easy to reproduce except when people are running things at heavy loads at scale and are correlating the data of crashes(not just chip failures). So guessing it’s simply if you load your chip often even at normal limits, it will start to become unstable kind of thing. The hotter the chip, the higher the rate of failure too. That’s my take from the analysis done.

What a disaster. How did they not stress test their chips for stability? Or worse, did they know and cover it up?

 

[CaptRiker]

SOOOOO glad last novemeber when I put together my first new system in 7 years ( was using an intel 5960x hedt system) I choose to go with an amd 7950x3d I choose amd due to much lower energy usage and alot less heat production. P H E W.

(have had ZERO issues with my new system, blazing fast and stable)

-----------------------------------
Fractal Design Meshify 2 RGB Black TG
AMD 7950x3d CPU (16c/32t) (pbo)
64 gigs ddr5-6000 CL30 Corsair Vengeance RGB Memory (expo)
Asus ProArt X670E-Creator Motherboard
(2.5+10gb Ethernet ports, 2x USB4 ports)
4tb Crucial T700 Gen5 ssd (12.4GB/sec read, 11.8GB/sec write) - Boot
4tb Samsung 990 Pro Gen4 ssd (7.4GB/sec read, 6.9GB/sec write) - Data
Gigabyte RTX 4080 GAMING OC 16G Graphics Card
Corsair H150i Elite Capellix XT 360mm AF120-RGB Water Cooler
1000watt CORSAIR HX Series HX1000i ATX 3.0
Windows 11 Home (22h2)
-----------------------------------

 

[TJ Hooker]

Hmm, 2 weeks ago he posted this on his LinkedIn page:



And now suddenly they have the ability to replace two Intel CPUs which "immediately exhibited instability" as well as "a few others" with unreleased AMD 9950Xs? Something's a bit sus...

A 7995WX system likely costs at least 5-10x as much as a 9950X system will. The latter is obviously much more affordable. Plus depending on what sort of warranty/replacement policy he has for his failing Intel systems, maybe he's able to get the AMDs systems as replacements rather than a full new purchase. Or maybe he just means that when he eventually needs to buy a new system going forward he will be looking at AMD over Intel. And as others have said, he could have been referring to a personal rather then work system with respect to the 7995WX.

There's nothing suspicious here, you're grasping at straws.

 

[Gururu]

Sounds very unprofessional to be ranting about business products on X. If this is as high level as it seems (are we talking about millions of dollars in failed chips?), this would be handled at an executive level between companies. Taking it to twitter sounds irresponsible and I’d have fired the weasel on the spot.

 

[CmdrShepard]

Are locked chips affected?

Very likely since they have the same microcode.

Do they need the microcode update?

Again very likely, but we will see which family / model / stepping will receive those updates.

How quick does degradation occur on those chips?

That heavily depends on usage pattern, cooling capacity, and previous default BIOS settings for PL1, PL2, and Vcore. Probably degrades faster if you have a lot of spiky single-threaded workloads such as compiling shaders in games.

 

[redgarl]

The story that never stop giving...

 

[HyperMatrix]

My 13900KS started degrading about 2 months after launch. Took months to get a replacement through warranty. This was the first time I was hit with degradation on intel since the 2600K. Generally I’d been able to push very high voltages and been fine as long as cooling was adequate. But the voltages on this chip are insane if you look at the predicted voltages in bios at stock clocks.

Edit: I should mention with the 13900KS after the first week of playing with the OC, I had gone back to running the chip at stock since I wasn’t able to get a meaningful OC out of the damn thing anyway. These chips are already pushed so hard to the edge that you’ll struggle to get even a 100-200MHz all-core OC out of it.

 

[ThomasKinsley]

Very likely since they have the same microcode.

I agree, although I'm hoping that the problem is caused from higher voltage spikes in unlocked chips with higher tdp and that degradation on locked chips is kept to a minimum. I'm about to set up a locked 14700 and I really want to make sure the chip is not being damaged just turning this thing on.

 

[Gururu]

I agree, although I'm hoping that the problem is caused from higher voltage spikes in unlocked chips with higher tdp and that degradation on locked chips is kept to a minimum. I'm about to set up a locked 14700 and I really want to make sure the chip is not being damaged just turning this thing on.

I've been running a 14700k for 6 months now but I didnt OC and dont run heavy loads for extended periods of time (WOW for many hours thats about it). Temps rarely over 70 C. So far impossible to know if anything has gone downhill but no signs. Hard to know what to do but got the 0x125 bios update just in case and will install the new one in August. Figure I've played it safe enough to not have had any accelerated degradation.

 

[tamalero]

Unfortunately the questions I have are the kind of questions that Intel does not want to answer. Are locked chips affected? Do they need the microcode update? How quick does degradation occur on those chips? When chips are breaking and the company is not answering questions, that does not induce confidence.

Their PR probably trying to salvage something to save this debacle.
Just like the crash of last friday.. I wonder how long until lawyers get involved..

 

[ThisIsMe]

Yeah, I think multiple problems are getting conflated due to the hype of everyone dogging on Intel and things will never get fixed this way.

Anyway, I have an i7 14700k system with an nVidia 4070. A few months or so ago, Unreal Engine released an update and shortly afterwards various games started releasing updates one after the other. It started with Fortnite. One day it worked, next day it updated and then always crashed when using direct X 12. Switch to DX 11 and everything is fine. Then other games began having issues, but only after updates. Always crash during DX 12 shader compilation/caching.

The crash happens in the nVidia driver. Just for fun I decided to throw in an AMD 7800XT in there to see what happens (I had in another system). And it works just fine. No crashes with either DX version.

Hard to explain that with this popular single issue theory. Everything else just works. Other games, compressing files, decompressing files, encryption, emulation, virtual machines, all work fine. System runs for weeks at a time without issues. Even when the UE games crash, the system remains fine afterwards.

 

[watzupken]

Sounds very unprofessional to be ranting about business products on X. If this is as high level as it seems (are we talking about millions of dollars in failed chips?), this would be handled at an executive level between companies. Taking it to twitter sounds irresponsible and I’d have fired the weasel on the spot.

I think we may not have the full picture to determine whether this is professional or not. It may be that they have tried to reach out to Intel and they denied them of repairs to replacements before. Hence they are taking it to social media to vent.
If you ask me, it’s equally unprofessional for Intel to openly blame it on motherboard makers when (1) they don’t enforce strict limits which I am sure they can, and, (2) they now figured the voltage applied is too high leading to failing chips. In all cases, their quest to retain performance crown at all cost, is now coming back to haunt them. You can floor the accelerator, but it only increases the risk of you crashing over time.

 

[TheHerald]

Yeah, I think multiple problems are getting conflated due to the hype of everyone dogging on Intel and things will never get fixed this way.

Anyway, I have an i7 4700k system with an nVidia 4070. A few months or so ago, Unreal Engine released an update and shortly afterwards various games started releasing updates one after the other. It started with Fortnite. One day it worked, next day it updated and then always crashed when using direct X 12. Switch to DX 11 and everything is fine. Then other games began having issues, but only after updates. Always crash during DX 12 shader compilation/caching.

The crash happens in the nVidia driver. Just for fun I decided to throw in an AMD 7800XT in there to see what happens (I had in another system). And it works just fine. No crashes with either DX version.

Hard to explain that with this popular single issue theory. Everything else just works. Other games, compressing files, decompressing files, encryption, emulation, virtual machines, all work fine. System runs for weeks at a time without issues. Even when the UE games crash, the system remains fine afterwards.

That's very hard to pinpoint. Changing your card fixed your crashing issues, but doesn't mean it wasn't the CPU. Maybe nvidia uses a heavier method of compiling shaders that results in crashes on unstable cpus. Troubleshooting crashes is a lot harder than troubleshooting...your socket being on fire.

 

[mitch074]

That's very hard to pinpoint. Changing your card fixed your crashing issues, but doesn't mean it wasn't the CPU. Maybe nvidia uses a heavier method of compiling shaders that results in crashes on unstable cpus. Troubleshooting crashes is a lot harder than troubleshooting...your socket being on fire.

There is also the fact that AMD drivers have lower CPU loads. If the crash is caused by a specific opcode that NVIDIA drivers call upon often to optimize DX12 code on an already strained compute unit, then a less strained compute unit using that opcode less often may not crash. It could be the same with an Intel GPU.