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

[dragontamer5788]

Alderon Games is some game developer for a game I don't play. But they came out with this post that has been going around various hardware circles on the internet: Intel is selling defective CPUs - Alderon Games

Anyone else hear more about this potential story? Two full generations of CPUs leading to stability issues? It looks quite devilish, it sounds like the CPUs ship out of Intel just fine but after months of use begin to have crashing / instability.

 

[Onasi]

Would probably be known by now if that was actually a thing. More cases would pop up. Intel DOES validate all its chips, selling defective products would be an insane move for them, the reputation losses are just not worth the potential monetary gain. So far, it seems that it’s the same instability nonsense caused by terrible OOB settings on motherboards that floated around couple months back. If that’s the case, Intel is less at fault of selling “faulty” chips and more for pushing them to the limit and then allowing MoBo vendors push them some more as a default.

 

[Super Firm Tofu]

There's obviously something going on with 13th/14th gen CPUs, but this sounds like the studio is bagging on intel as free marketing. At least I've never heard of them or their games. :shrug:

 

[dragontamer5788]

Would probably be known by now if that was actually a thing. More cases would pop up. Intel DOES validate all its chips, selling defective products would be an insane move for them, the reputation losses are just not worth the potential monetary gain. So far, it seems that it’s the same instability nonsense caused by terrible OOB settings on motherboards that floated around couple months back. If that’s the case, Intel is less at fault of selling “faulty” chips and more for pushing them to the limit and then allowing MoBo vendors push them some more as a default.

People don't overclock servers though.

Consumer devices? Sure, I can see that being a problem. But this game studio is saying that they've seen similar problems in servers (which often are clocked lower for more long term stability)

There's obviously something going on with 13th/14th gen CPUs, but this sounds like the studio is bagging on intel as free marketing. At least I've never heard of them or their games. :shrug:

The thought has certainly crossed my mind that this is just an attempt at free marketing by making a mountain out of a molehill.

 

[Steevo]

Its a combination of selling chips that are running on the bleeding edge of their voltage curve, and motherboard manufacturers having aggressive boost profiles... that was encouraged by Intel, and recently Intel threw the board makers under the bus as chips were degrading, faulting and failing when supplied the aggressive voltage and boost frequency that Intel documents suggested were OK to use

 

[Super Firm Tofu]

Its a combination of selling chips that are running on the bleeding edge of their voltage curve, and motherboard manufacturers having aggressive boost profiles... that was encouraged by Intel, and recently Intel threw the board makers under the bus as chips were degrading, faulting and failing when supplied the aggressive voltage and boost frequency that Intel documents suggested were OK to use

Take a look at the video I posted above. It's the first non-drama look at the issue that points to more than just being pushed too far.

 

[eidairaman1]

Its a combination of selling chips that are running on the bleeding edge of their voltage curve, and motherboard manufacturers having aggressive boost profiles... that was encouraged by Intel, and recently Intel threw the board makers under the bus as chips were degrading, faulting and failing when supplied the aggressive voltage and boost frequency that Intel documents suggested were OK to use

Just like ford and firestone, in it to blame eachother instead of fixing it

 

[64K]

Only thing I know about Alderon Games is that they have a MMO game (Path of Titans) that's been in Beta for 4 years now. Maybe they are trying to drum up some free publicity to get cash to finally finish the game?

 

[eidairaman1]

There's obviously something going on with 13th/14th gen CPUs, but this sounds like the studio is bagging on intel as free marketing. At least I've never heard of them or their games. :shrug:

Sounds like a pch issue

 

[Onasi]

People don't overclock servers though.

Consumer devices? Sure, I can see that being a problem. But this game studio is saying that they've seen similar problems in servers (which often are clocked lower for more long term stability)

Server chips don’t even belong to the same 13/14th gen they are talking about. They are distinctly different with Intel. So either the studio is bullshitting, of they are running consumer chips for their servers somehow, which is a terrible idea anyway.

 

[Super Firm Tofu]

Server chips don’t even belong to the same 13/14th gen they are talking about. They are distinctly different with Intel. So either the studio is bullshitting, of they are running consumer chips for their servers somehow, which is a terrible idea anyway.

In this case they are. 13900k/14900k on the W680 chipset. (at least in Level1Tech's video)

 

[Onasi]

In this case they are. 13900k/14900k on the W680 chipset.

Well, there you have it. Neither the CPUs, nor the chipset are intended for this use. Workstations - sure. Using them for hosting dedicated servers for an MMO of all things is just stupid. Don’t get me wrong, I have issues with how far Intel pushed the i9 models these past couple generations, it absolutely can potentially lead to instability if used improperly, but this is very much a case of user error too.

 

[oxrufiioxo]

I haven't watched it becuase honestly I don't care much for 13900k/14900k cpus to begin with but it has been mildly interesting to see what happens when a cpu gets pushed to the edge and board makers do whatever they want with voltages like they've done at least since the early 2010s and probably before that. I honestly can't remember a single intel board I've owned over the past 15 years that didn't set dubious voltages at stock with XMP enabled I've had to manually tune them all, now would the cpu actually degrade if I didn't who knows but I'm not taking any chances.... it started getting really bad during 9th gen for me on asus boards to the point even the board would mark the voltages as Red lol.

Not saying that is what's going on with 13/14th gen i9s but I am not surprised we've finally come to this point.

 

[GenericNinja]

Well, there you have it. Neither the CPUs, nor the chipset are intended for this use. Workstations - sure. Using them for hosting dedicated servers for an MMO of all things is just stupid. Don’t get me wrong, I have issues with how far Intel pushed the i9 models these past couple generations, it absolutely can potentially lead to instability if used improperly, but this is very much a case of user error too.

Intel is selling directly to these providers. They are even throwing trays of free CPUs at them according to the videos to try to mitigate this problem without a current solution. Using W680 and I9 for a game server which needs relatively few very high performance threads is a perfectly appropriate use case, not user error. Game server companies have been doing this for years and it wasn't a problem for AMD. It wasn't a problem for Intel previously, up to 12th gen was/is fine. 13th and 14th are the problem.

 

[Dr. Dro]

Server chips don’t even belong to the same 13/14th gen they are talking about. They are distinctly different with Intel. So either the studio is bullshitting, of they are running consumer chips for their servers somehow, which is a terrible idea anyway.

Equivalent Xeon E-class CPUs exist, for example, the Xeon E-2488 is a i9-13900K without E-core clusters enabled. Anyhow... I don't think it's defensible at this point. Intel needs to do a mass recall for the 13th and 14th generation Core i9 K, KF and KS processors, for a full refund, or release a fix pronto. We've given them enough time. This recall should be elective, if an user feels like returning their chip because they don't trust it - no questions asked.

I haven't watched it becuase honestly I don't care much for 13900k/14900k cpus to begin with but it has been mildly interesting to see what happens when a cpu gets pushed to the edge and board makers do whatever they want with voltages like they've done at least since the early 2010s and probably before that. I honestly can't remember a single intel board I've owned over the past 15 years that didn't set dubious voltages at stock with XMP enabled I've had to manually tune them all, now would the cpu actually degrade if I didn't who knows but I'm not taking any chances.... it started getting really bad during 9th gen for me on asus boards to the point even the board would mark the voltages as Red lol.

Not saying that is what's going on with 13/14th gen i9s but I am not surprised we've finally come to this point.

I watched it. Wendell talks about how the chips are behaving under Linux and an approximate failure rate of 50%. It seems to be caused by extreme degradation of the processor in a short to medium term operation, the processor will operate error free for a number of weeks (sometimes up to 2 or 3 months), and begin to exhibit symptoms over time. It gives off timebomb vibes. The CPUs are failing on machines with the workstation version of the Alder Point chipset (W680), which is just a Z690 with a multiplier lock, vPro and ECC support. I consider myself pretty lucky that my ride has been mostly bump free with the 13900KS...

 

[dragontamer5788]

I consider myself pretty lucky that my ride has been mostly bump free with the 13900KS...

I actually had a bunch of those 3TB Seagates in a NAS way back when they were having 20%+ error rates each year. None of my drives ever failed despite matching exactly the model that everyone complained about.

Sometimes you just get lucky. But yeah, weird issues like this are always so hard to debug, because so many users also just have no problems with it.

 

[oxrufiioxo]

Equivalent Xeon E-class CPUs exist, for example, the Xeon E-2488 is a i9-13900K without E-core clusters enabled. Anyhow... I don't think it's defensible at this point. Intel needs to do a mass recall for the 13th and 14th generation Core i9 K, KF and KS processors, for a full refund, or release a fix pronto. We've given them enough time. This recall should be elective, if an user feels like returning their chip because they don't trust it - no questions asked.



I watched it. Wendell talks about how the chips are behaving under Linux and an approximate failure rate of 50%. It seems to be caused by extreme degradation of the processor in a short to medium term operation, the processor will operate error free for a number of weeks (sometimes up to 2 or 3 months), and begin to exhibit symptoms over time. It gives off timebomb vibes. The CPUs are failing on machines with the workstation version of the Alder Point chipset (W680), which is just a Z690 with a multiplier lock, vPro and ECC support. I consider myself pretty lucky that my ride has been mostly bump free with the 13900KS...

At first I thought it was a repeat of the 4090 12VHPWR connector thing with a very small amount of users either by error or poor tolerances on the connector in QC but this seems like a much bigger issue.

I was looking at a 13900k before going am5 could have gotten a whole setup for 600 usd with a semi decent board but didn't want to test my luck I likely would have locked it around 140w same as my 7950X3D anyways so it probably wouldn't have mattered. Looking at gaming performance at 125w on a 14900k is mostly what made me go nah and that was locally not benchmarks.

 

[thesmokingman]

There's obviously something going on with 13th/14th gen CPUs, but this sounds like the studio is bagging on intel as free marketing. At least I've never heard of them or their games. :shrug:

Did you watch the video actually?? Like dude, there's a reason data center support cost on those chips is almost 10x their AMD counterpart.

 

[dragontamer5788]

At first I thought it was a repeat of the 4090 12VHPWR connector thing with a very small amount of users either by error or poor tolerances on the connector in QC but this seems like a much bigger issue.

12VHPWR is very reminiscent of this.

When you solder a connection like that, its reliable out of the factory. But in real life, wires move, especially as users try to squeeze wires into place. Solder is very inflexible, lead is brittle and it forms internal fractures easily. Especially because the copper wires are both stronger and more flexible. So the copper moves and squeezes and bends into shape, while the lead solder snaps, breaks, and falls apart internally.

Its neither user error nor bad factory testing. Its just knowledge about what happens in the long term and someone somewhere being ignorant of long-term issues. It only took one manufacturer to make leaded / soldered on wires on the 12VHPWR lines for it all to go to crap (and everyone who performed testing from all angles would have seen things perfectly fine in the factory).

Something like that must be going on: maybe internal glue, thermal paste, mechanical stresses, heat stresses... there's a lot of issues when chips get hot (and cold), thermal cycling. There's electrical stresses involved: low voltages but very high currents. Those also could cause long term reliability issues.

----------

Do we even know if its the CPU itself that's the problem yet? It could be the northbridge, it could be the voltage regulators / VRMs. It could be inductors or capacitors.

 

[oxrufiioxo]

12VHPWR is very reminiscent of this.

When you solder a connection like that, its reliable out of the factory. But in real life, wires move, especially as users try to squeeze wires into place. Solder is very inflexible, lead is brittle and it forms internal fractures easily. Especially because the copper wires are both stronger and more flexible. So the copper moves and squeezes and bends into shape, while the lead solder snaps, breaks, and falls apart internally.

Its neither user error nor bad factory testing. Its just knowledge about what happens in the long term and someone somewhere being ignorant of long-term issues. It only took one manufacturer to make leaded / soldered on wires on the 12VHPWR lines for it all to go to crap (and everyone who performed testing from all angles would have seen things perfectly fine in the factory).

Something like that must be going on: maybe internal glue, thermal paste, mechanical stresses, heat stresses... there's a lot of issues when chips get hot (and cold), thermal cycling. There's electrical stresses involved: low voltages but very high currents. Those also could cause long term reliability issues.

----------

Do we even know if its the CPU itself that's the problem yet? It could be the northbridge, it could be the voltage regulators / VRMs. It could be inductors or capacitors.

Yeah, all 5 people I know personally with 4090s haven't experienced a connector melting although none use one of those recalled cablemod angle adapters either all first generation connectors but 2 out of 5 people I know with 13900k/14900k have expierenced this issue with stock settings so it seems more widespread probably too small of a sample size though. One of them tried to rma and Intel was like nah dog but thankfully amazon replaced it.

I actually tried really hard to get my Gigabyte 4090 to melt couldn't do it even with the connector barely seated really strange issues for sure.

 

[Super Firm Tofu]

Did you watch the video actually?? Like dude, there's a reason data center support cost on those chips is almost 10x their AMD counterpart.

Uh, yeah 'dude', I did, yesterday. Just because there's a big problem with 13th/14th gen right now that doesn't appear to even have a known cause, let alone a fix, doesn't mean that no-name studio with no-name beta MMO isn't doing this for the publicity. I mean everyone is talking about it on the tech sites right now, right?

 

[thesmokingman]

Uh, yeah 'dude', I did, yesterday. Just because there's a big problem with 13th/14th gen right now that doesn't appear to even have a known cause, let alone a fix, doesn't mean that no-name studio with no-name beta MMO isn't doing this for the publicity. I mean everyone is talking about it on the tech sites right now, right?

There's clearly a serious issue and you took that opportunity to suggest its not and this dev is bagging on Intel for marketing. Shill much?

 

[Super Firm Tofu]

There's clearly a serious issue and you took that opportunity to suggest its not and this dev is bagging on Intel for marketing. Shill much?

*Sigh*. In neither of my posts did I suggest that there's no problem. What are you so mad about? The problem with 13th/14th gen and the dev self-promoting can both be true.

 

[ir_cow]

I'm going to put my bet down and say the System Agent voltage is "killing" these CPUs.

 

[thesmokingman]

*Sigh*. In neither of my posts did I suggest that there's no problem. What are you so mad about? The problem with 13th/14th gen and the dev self-promoting can both be true.

You're posts are contradictory. Go read your post again lmao.

There's obviously something going on with 13th/14th gen CPUs, but this sounds like the studio is bagging on intel as free marketing.

So is there a problem or not? Why's it free marketing if there is a problem? What is shit?