Alderon Games claims that substantial numbers of Intel 13th Gen and 14th Gen chips are defective

[Klemc]

• There is an issue with certain newer-gen Intel CPUs crashing on startup for many UE5 games, including ours. This appears to be a hardware issue and the only workaround is to modify BIOS settings. More information here: https://cyan.com/docs/riven-launche...to-the-menu-sometimes-shows-an-error-message/

https://steamdb.info/patchnotes/15016294/

After all the laughs of the 7800x3d thingy !!!

 

[RJARRRPCGP]

Of course CPU's all degrade although usually its not a visible thing on chips that run at stock, I have seen a top end AMD FX chip being no longer stable at stock, that was in AMD's dark days where they were pushing chips hard.

The FX 9590s have a much stronger tendency to fail! My buddy's FX 9590 failed extremely fast! Seems that FX 8350s were reliable, OTOH. I didn't even have to increase the Vcore for 4.4 GHz on mine (date code 14xx) with x.264, before giving it away.

In fact, the 13th-gen and 14th-gen, starkly reminds me of my buddy's FX 9590! IIRC, there were other reports of FX 9590s failing, too!

FFS, he don't even OC! And, he couldn't even Skype without his PC going down! (That incident was in 2014, which was before we switched to Discord, when it was the Hammer&Chisel era!)
(Which we switched to, by 2017, possibly just within some months later in 2016) (We were still using Skype in 2015)

 

[Steevo]

https://www.epicgames.com/help/en-US/c-Category_Fortnite/c-Fortnite_TechnicalSupport/frequent-crashes-in-fortnite-on-i9-13900k-kf-ks-or-i9-14900k-kf-ks-cpus-a000086852?sessionInvalidated=true

Fortnite has joined the fray...

 

[InVasMani]

Imagine buying a Apex to run at 150w TPD. Sad times.

On the topic of cache/ring. I have mine locked at 49x. It becomes very unstable at 52x+. Maybe my CPUs is a buggy one but it just happens to I lock all the voltages and multiplier. Some of those system crashes level1 was describing are signs are unstable core / voltages.

I've seen memory errors even at x40 ring that went away at x39 ring depending on MT/s and cas of my memory. I started at x8 ring and bumped it up slowly over time and ran x86 memtest a good period. It's once I bumped up to x40 or higher I began seeing issues with like hammer row tests and like inversion tests pop up.

It'll run higher, but you'll start seeing a lot more errors at the same memory timings and voltages is what I had noticed. I found x40 about the upper limits and sort of tolerable/intolerable break point at 7000MT/s CL30. That said if I loosen up CAS and/or drop MT/s you've got more leeway with the ring I believe.

I use below 1v on SA it doesn't seem to have a readily noticeable impact on overall stability for me below a certain point you'll run into bios post issues, but otherwise not much concern setting it lower. That said it's pain if you set it too low because you'll end up in a boot loop cycle and need to clear CMOS.

So like be advised it's pain in the neck annoyance even with a clear CMOS switch it's a general aggravation honestly when you get stuck in limbo land seeing if it'll post or just loop and struggle and eventually you get tired and clear the CMOS. Like if you've been there you'll understand waiting on a bios to maybe bios or maybe decide it doesn't want to it's painfully slow sometimes.

Anyway it seems to me suspect of ring ratio in relationship to L cache voltage and overall general memory times though primarily MT/s and CAS given those are the two most obvious ones to start with command rate, gear ratio, and others of course also play a role.

I feel like ultimately the cache in relationship to voltage and memory alongside the ring just hits a cave in point in reliability and performance expectations around them all. The cache voltage and ring impacts the x86 memtest block level cache performance pretty readily relative to memory settings. Sometimes it's one thing or another that generally limits the overall block level cache test performance results significantly or marginally. I settled on x40 ring because I live dangerously sometimes threading the needle system stability in favor of hint of additional performance. I'm just a little OCD for system instability it really has nothing to do with performance.

I think part of the reason we're hearing a lot about the 13900K and 14900K chips in particular is they've got more heat output of course, but also they've got more CPU cores on the ring bus which ultimately is additional stress to the ring bus. Plus consumers are like hey it's better silicone lottery thus I can probably push the IMC harder which piles on more stress to the ring bus. It's K model I can probably OCer it a bit harder because binning. Which once again piles on more stress to the ring bus. Like the ring bus is just being dumped on heavily so that's likely no small part of it.

I think there a number of general faults to go around though the fact that server chips are failing is really bad indicator without a shadow of doubt. Like irrespective of consumers OCers and those scenario's these chips are failing on their own under normal conditions and just heavy continuous usage that is expected of them hold up for a number of reasonable years perfectly fine. Given that fact it looks very much like a Intel problem.

I get why they pointed the finger at MB makers those issues didn't exactly help matters. They also were serious enough concerns and faults on MB makers part. Like I'm fine with both Intel and AMD scolding MB maker's being stupid about dubious default configuration settings. Like they gotta stop doing that irrespective of this Intel issue or the one AMD had with X3D chips and voltages and frying that was at least more readily fixable from looks and sounds of it.

Intel needs to suck it up and do the right thing already especially if their well aware of the the root of the problem. You can't just beg tax payers for billions in contributions then turn around and screw over many of those same tax payers at the same time. It's not a scenario you walk away from unscathed that's literally corporate suicide. Build your foundries all you like, but no one is going to trust your products if you screw this situation up. At least the same time other companies aren't going to place much faith in your foundries if you screw this up heavily. Like that doesn't build trust and confidence. They need own up to it if they know what's happening and address it properly.

 

[lilhasselhoffer]

....

Intel needs to suck it up and do the right thing already especially if their well aware of the the root of the problem. You can't just beg tax payers for billions in contributions then turn around and screw over many of those same tax payers at the same time. It's not a scenario you walk away from unscathed that's literally corporate suicide. Build your foundries all you like, but no one is going to trust your products if you screw this situation up. At least the same time other companies aren't going to place much faith in your foundries if you screw this up heavily. Like that doesn't build trust and confidence. They need own up to it if they know what's happening and address it properly.

Billions in bonuses amongst bank bailouts


Couldn't, shouldn't, and can't are all different things.
Intel and AMD are being courted to make chips in the US, so Taiwan is less of a global spark point. The US is doing this with monetary incentives, that do not have strings. This is therefore not a "bad" thing for Intel unless it becomes a mainstream issue...and most mainstream outlets will fumble the ball trying to explain internal overclocking versus external overclocks.


My two cents is that Intel has been running without innovation for so long that their "give it more gas" methodology of improving performance is finally catching up...and I don't like where AMD's 7xxx series went with the same logic either. It's burning a chip very bright for little gains...and here I am sitting on some 3770k units that only now refuse to clock at anything but stock after a decade of usage. To be fair though, 10 years under water at only mild (voltage boosts to maintain all slot memory stability) overclocking is pretty much a joke....but I'm hoping my current 5xxx series AM4 platform has the same longevity with less compromises.

 

[Assimilator]

Intel and AMD are being courted to make chips in the US, so Taiwan is less of a global spark point. The US is doing this with monetary incentives, that do not have strings.

AMD has been fabless for a decade and a half, Intel's nodes are so shit that they're turning to TSMC, and TSMC - being majority owned by the Taiwan government - is never going to build foundries encapsulating its cutting-edge nodes anywhere other than Taiwan, because if it does the island loses its anti-PRC defence in the form of the USA.

Now of course, for commodity semiconductors that can be fabricated on older and cheaper nodes, none of the above is relevant. But even then, and even with the incentives being handed out by the West, it's still cheaper to produce those semiconductors in the East and ship them halfway around the world, due to labour costs (Vietnam is a strong prospect in this regard). Even if that calculus changes - i.e. if the PRC invades Taiwan in this decade, as many suspect, and the West dedicates war-level funds to obtaining TSMC-level nodes locally - constructing a foundry is not a quick affair, and a TSMC-level foundry especially not.

 

[Darmok N Jalad]

The most likely reason that devs are pointing the finger at Intel and making a big deal out of this is because when their stuff crashes, they are the ones who have to figure out what is going on, and then they have to try to fix it. They get to do the work, they get the complaints. It’s a bit much to assume they are doing this “for attention,” especially if several devs are raising the flag. No, they are likely tired of dealing with the issue and have no other recourse. There’s probably a really good chance that any studio worth anything is getting early access to the hardware so they can validate. Do you really think they want to stir the pot with a giant hardware company? No, I think they are fed up testing and retesting. We saw the same with NVIDIA, they got tired of the crash reports too and pointed at motherboard vendors, who then aimed things back to Intel. And then Intel has acknowledged some issues that they can’t fully resolve. Too much evidence to say that there’s a real issue—one that doesn’t seem easily fixed.

The other concern is that if this does get patched and worked out, it is likely going to come at a performance penalty. It won’t matter as much by then, since these models will get EOLed and something else will replace them. There will probably be a class action lawsuit here at the end that is settled as quickly as possible. It would be one thing if it was just random users complaining on the internet, but it’s random users AND studios AND hardware companies AND even Intel saying something’s up. Too much smoke to not find a fire.

 

[thesmokingman]

Guys stop feeding the shill.

 

[Dr. Dro]

My 13900KFC got a one day pass from Fortuneteller Baba after it died. Like every great warrior, it braved Snake Way and trained with King Kai himself.

...you people. Does a problem exist? Yes. Did every single chip ever produced die? No. Will every single chip that has ever been deployed malfunction? Unlikely. Intel will get their issue sorted, there's no need for such doom and gloom, nor does wishing any misfortune upon Intel or other people benefit you - even if you're an AMD diehard - at all. Especially you AMD diehards that happen to be value-conscious and like them because of monetary reasons: if they gain an ample lead, you are gonna feel it in your wallet.

 

[Vario]

I just find it odd that the 12900K is spared this fate.

 

[Solaris17]

This thread isnt about your AMD vs. Intel crap or an excuse to bring out random benchmark numbers for 0 reason.

Stick to the topic thanks.

 

[Outback Bronze]

We all win when both companies are winning in their own way.

Yeah, for this particular thread I'd say I'm wining as I've shown that I'm 4700 hours + with no instability issues, but get ridiculed the way I want to run my processor...

I've done all the tests I need to do for this cpu and 1700 combo and have retired it @ 4.5ghz. I see no personal need to run this processor like a bat out of hell 300w+ putting extra strain on all my rails and my cooling system. Maybe this is half the problem? I have also got 16 E core's at my disposal if I ever need extra threading. I pretty much run the 65w profile that wizard has tested and it's got good all round performance.

I'm really looking forward to AMD's release in the next couple of weeks, then see what arrow lake has to offer and now Intel's only P core CPU's on 1700.

 

[oxrufiioxo]

Yeah, for this particular thread I'd say I'm wining as I've shown that I'm 4700 hours + with no instability issues, but get ridiculed the way I want to run my processor...

I've done all the tests I need to do for this cpu and 1700 combo and have retired it @ 4.5ghz. I see no personal need to run this processor like a bat out of hell 300w+ putting extra strain on all my rails and my cooling system. Maybe this is half the problem? I have also got 16 E core's at my disposal if I ever need extra threading. I pretty much run the 65w profile that wizard has tested and it's got good all round performance.

I'm really looking forward to AMD's release in the next couple of weeks, then see what arrow lake has to offer and now Intel's only P core CPU's on 1700.

The 14900k and really the 13900k are extremely impressive cpu's anyone who thinks otherwise are just being delusional in their fanboyism and it's up to the user to choose how they run them. Intel just needs to take care of those affected and move on regardless if that is 5% or 50% of people who purchased them.

There are enough reports about this to know something is going on how widespread it is who can say but I know if I was affected I would want the company to man up and make it right.

 

[Darmok N Jalad]

Yeah, for this particular thread I'd say I'm wining as I've shown that I'm 4700 hours + with no instability issues, but get ridiculed the way I want to run my processor...

I've done all the tests I need to do for this cpu and 1700 combo and have retired it @ 4.5ghz. I see no personal need to run this processor like a bat out of hell 300w+ putting extra strain on all my rails and my cooling system. Maybe this is half the problem? I have also got 16 E core's at my disposal if I ever need extra threading. I pretty much run the 65w profile that wizard has tested and it's got good all round performance.

I'm really looking forward to AMD's release in the next couple of weeks, then see what arrow lake has to offer and now Intel's only P core CPU's on 1700.

Personally, I'm not ridiculing you one bit. The problem in your situation is that you are not at all running your processor the way Intel and motherboard makers have configured it out of the box. Your situation really only adds more evidence that suggests there is indeed a problem with running these chips at this really vague "default configuration." The alleged problem is with the highest SKUs, which run at the most extreme voltage and thermal curves. You have effectively run your chip at a much safer operating condition, with power demands more similar to the lesser SKUs that are reportedly not having any issues. The speculation is that these highest SKUs are simply being pushed way too hard, causing failure.

We can certainly argue that enthusiast/extreme-class CPUs require more refinement than what the factory defaults provide, and hats off to you for taking the time to properly configure your chip. The suspected problem is the idea that the default settings appear to be unstable and likely detrimental to the processor itself. This simply shouldn't be. These chips should default-boot to conservative/safe operating conditions and then it's on the user to push past that. That's not what we appear to be seeing, and between motherboard makers aggressively "auto-overclocking" and Intel not policing it, we ultimately get a plug-n-play lemon. The chips have great potential, but the entire situation is ugly, and no one wants to take the blame. Just my opinion, but I think Intel took a blind eye to motherboard makers, and motherboard makers had their way with these chips, I guess because the average user is simply too lazy to enter the BIOS and configure things themselves.

Personally, I just want a plug-n-play setup anymore. I want low-fuss, and default settings that are safe enough to not destroy the hardware. Knowing my preferences and limits, I'd never buy a top SKU though. The mid-range seems to offer the best compromise of performance, stability, and efficiency with low-effort. Then again, I'm running a 2667v2, but that's how I'd buy today if I wanted to.

 

[Panther_Seraphin]

I just find it odd that the 12900K is spared this fate.

Spoiler: WRONG

12900k was still on 14nm process node where as the 13th gen onwards are on Intels 10nm node so its probably something in the process that is degrading/failing at these higher voltages.

It is on Intel 7, I was looking at Rocket lake/11th gen when I read it was on 14nm

Intel and 10nm havent been best of buds with their initial 10nm parts being slower than an equivalent 14nm part at the time. Add to the fact that they are jumping over to TSMC to manufacturer more and more of their tiles/dies makes me worry a little in the short term for Intel and their foundaries. I honestly hope they get the jump on people with EUV tech so it forces TSMC to keep pushing and hopefully to lower their prices with the competition.

Lets hope we can just look back at this node like a cursed time for intel similar to the P3 1Ghz chip/end of Netburst/Itanium sort of "nostalgia"

 

[Klemc]

12900k was still on 14nm process node where as the 13th gen onwards are on Intels 10nm node so its probably something in the process that is degrading/failing at these higher voltages.

Intel and 10nm havent been best of buds with their initial 10nm parts being slower than an equivalent 14nm part at the time. Add to the fact that they are jumping over to TSMC to manufacturer more and more of their tiles/dies makes me worry a little in the short term for Intel and their foundaries. I honestly hope they get the jump on people with EUV tech so it forces TSMC to keep pushing and hopefully to lower their prices with the competition.

Lets hope we can just look back at this node like a cursed time for intel similar to the P3 1Ghz chip/end of Netburst/Itanium sort of "nostalgia"

So that's why the 10nm got delayed years...... OKay.

 

[londiste]

12900k was still on 14nm process node where as the 13th gen onwards are on Intels 10nm node so its probably something in the process that is degrading/failing at these higher voltages.

Both 12th and 13/14th gen are on Intel 7 (previously known as 10nm).

 

[RJARRRPCGP]

I pretty much run the 65w profile that wizard has tested and it's got good all round performance.

Like with Vermeer, I think it will do well in gaming performance. It's the maximum-multicore tasks that are using the most watts. So no surprise if gaming performance is excellent, unless the game needs so many cores. It's not all about "more cores", as you observed.

 

[eidairaman1]

The FX 9590s have a much stronger tendency to fail! My buddy's FX 9590 failed extremely fast! Seems that FX 8350s were reliable, OTOH. I didn't even have to increase the Vcore for 4.4 GHz on mine (date code 14xx) with x.264, before giving it away.

In fact, the 13th-gen and 14th-gen, starkly reminds me of my buddy's FX 9590! IIRC, there were other reports of FX 9590s failing, too!

FFS, he don't even OC! And, he couldn't even Skype without his PC going down! (That incident was in 2014, which was before we switched to Discord, when it was the Hammer&Chisel era!)
(Which we switched to, by 2017, possibly just within some months later in 2016) (We were still using Skype in 2015)

Those were 220W TDP Chips for toying with.

Vs a 125W TDP Chip.

 

[thesmokingman]

It keeps expanding, now Alderon is saying...

"Yes we have several laptops that have failed with the same crashes. It's just slightly more rare then the desktop CPU faults," the dev posted.

Alderon Games claimed several days ago that virtually 100% of its Raptor Lake [desktop] chips were crashing, so the fact that Raptor Lake laptop crashing is only "slightly more rare" reveals how truly problematic the issue is on all platforms.

Dev reports Intel's laptop CPUs are also suffering from crashing issues — several laptops have suffered similar failures in testing

New evidence busts any theories that Raptor Lake instability is a desktop-specific issue.

www.tomshardware.com

 

[Vario]

It keeps expanding, now Alderon is saying...

Dev reports Intel's laptop CPUs are also suffering from crashing issues — several laptops have suffered similar failures in testing

New evidence busts any theories that Raptor Lake instability is a desktop-specific issue.

www.tomshardware.com

Maybe its just their code.

 

[Caring1]

Steve from GN just released a video not long ago, and it's not just code that is the issue with 13th and 14th Gen.

 

[Taraquin]

I wonder how stats on 13600K is?

 

[londiste]

Steve from GN just released a video not long ago, and it's not just code that is the issue with 13th and 14th Gen.

tl;dw
Multiple and believable leaks/tips say Intel has or had a manufacturing issue that leads to copper via oxidation on Raptor Lake CPUs.

 

[Panther_Seraphin]

tl;dw
Multiple and believable leaks/tips say Intel has or had a manufacturing issue that leads to copper via oxidation on Raptor Lake CPUs.

If confirmed true.....God Damn this is gonna be expensive for intel as basically all 13 and 14th gen CPUs are going to die