Intel Will Not Recall Failing 13th and 14th Gen CPUs

[kawice]

People running stock intel profile have issues. Server motherboards that have zero OC and even have lowered memory speeds have issues. Laptops have issues. Even low power T series chips are having issues. It's irrelevant how many users are facing this problem due to overclock because right now it seems to be a matter of luck more than anything else. Besides the number of users OC'ing these days is very small regardless.

And what does contact frame have to do with this? Contact frame improves the mounting pressure and temps but this issues does not seem temperature related based on Wendell's data as none of the server CPU's exceeded 83c. If ESD occurred it would kill the motherboard first, not the CPU.

The issue showcased months after 13th gen release, so it might affect some batch of CPUs. Intel Profile in BIOS is just a patch for degraded CPUs. You'll be better off using optimized profile that doesn't boost CPU voltage over the top to "stabilize" damaged CPUs.

The contact frame is just a "snake-oil" product and only useful for extreme overclockers. It's usefulness was debunked by GamerNexus.
The contact frame can introduce instability itself if not ideally mounted, which is difficult to do in home conditions.

AFAIK, ESD can kill any electronic device, doesn't mater if it's motherboard, CPU , RAM, or GPU. The voltage is high enough to do damage.

 

[Dr. Dro]

Could all of this be the reason why we are getting a 1700 CPU with just P cores and n HT. I think it is called Bartlett. As much as some people think that this is being inflated did not watch Wendlell's video completely.

Bartlett has nothing to do with this. Rumors of it have been circulating since last year. It's supposedly being developed for/by the Intel NEX division, but a DIY channel release has always been rumored. Even if it remains being a Raptor Cove design (it likely is since it's rumored to support HT), as long a this new processor supports AVX-512 and perhaps a comes with an integrated NPU, it'll be a clear cut upgrade path from Raptor Lake on the same socket.

 

[kapone32]

Bartlett has nothing to do with this. Rumors of it have been circulating since last year. It's supposedly being developed for/by the Intel NEX division, but a DIY channel release has always been rumored. Even if it remains being a Raptor Cove design (it likely is since it's rumored to support HT), as long a this new processor supports AVX-512 and perhaps a comes with an integrated NPU, it'll be a clear cut upgrade path from Raptor Lake on the same socket.

Sorry no AVX512 support. How would it be a better CPU if there are less cores and no HT support which has been around for at least 10 years. The fact remains that there is a flaw with the current CPUs, even a 5% failure rate would be unacceptable. The data is there as there is more than one user who has confirmed that there is something wrong. You can believe what you want but only Intel know how long this issue has existed.

There is also the huge elephant in the room that by the time these launch the new X3D chips should be in the channel. I guess it makes sense though if you want to have stability or your CPU has degraded by the time these launch so that the expense is not as great. This makes them not recalling these chips make more sense in a world where the biggest tech tubers are now hated by Intel fans for exposing the truth.

 

[Sound_Card]

There is also the huge elephant in the room that by the time these launch the new X3D chips should be in the channel. I guess it makes sense though if you want to have stability or your CPU has degraded by the time these launch so that the expense is not as great. This makes them not recalling these chips make more sense in a world where the biggest tech tubers are now hated by Intel fans for exposing the truth.

I don't think there is really a true intel fan. At least when it comes to the Zen era, I think the vast majority of people who buy Intel are actually rabid Nvidia fans. They typically have an 80 class or 90 class GPU too. I mean this in the high end CPU segment. I believe the 10400(k) was a great mid range CPU for the money, perhaps offering better value than AMD.

 

[Tomorrow]

It's really hard to trust the techtubers on this one because up until now they haven't earned my trust. Intel has.

If they haven't earned your trust by now by exposing and investigating countless anti-competitive and anti-consumer developments then i dont know what will.
I will NEVER trust any corporation over them. At least some of the bigger tech tubers. Obviously i cant speak for every smaller one.

The cope is intense, and i'm not surprised given how much mindshare Intel has. Trusting intel over reputable tech youtubers bringing factual information here? Without tech youtubers bringing forth the corrosion issue, and the rapid degradation of i9's I doubt intel would have even admitted the 13th gen cpus had a degradation problem. Sure there are a bunch of people jumping on the topic just repeating what GN and Wendell has said but that doesn't dismiss cpus failing being a problem.

Imagine trusting Intel (of all corporations) over people who do this as a living and have their viewers, instead of sponsors and shareholders as income source.

The contact frame is just a "snake-oil" product and only useful for extreme overclockers. It's usefulness was debunked by GamerNexus.
The contact frame can introduce instability itself if not ideally mounted, which is difficult to do in home conditions.

AFAIK, ESD can kill any electronic device, doesn't mater if it's motherboard, CPU , RAM, or GPU. The voltage is high enough to do damage.

The contact frame is a solution to Intel's shoddy engineering. It offers real benefits in terms of temperatures. GN did not debunk it's usefulness but the implementation matters and not all contact frame options are created equal. I fail to see how mounting it in home conditions is difficult. People do far more complex stuff of delidding, using liquid metal etc and unless it's user error then it's mostly fine.
ESD risk these days sis pretty small unless the user literally puts the motherboard on a carpet or runs the system without a case.
Proper work mats and cases are cheap these days.

Bartlett has nothing to do with this. Rumors of it have been circulating since last year. It's supposedly being developed for/by the Intel NEX division, but a DIY channel release has always been rumored. Even if it remains being a Raptor Cove design (it likely is since it's rumored to support HT), as long a this new processor supports AVX-512 and perhaps a comes with an integrated NPU, it'll be a clear cut upgrade path from Raptor Lake on the same socket.

Last year when Intel internally likely already knew that 13th gen were failing? maybe.
And it wont have AVX-512 because that's a workstation upsell feature now nor an useless NPU that no one has figured out a good use case yet.

 

[Dr. Dro]

Sorry no AVX512 support. How would it be a better CPU if there are less cores and no HT support which has been around for at least 10 years. The fact remains that there is a flaw with the current CPUs, even a 5% failure rate would be unacceptable. The data is there as there is more than one user who has confirmed that there is something wrong. You can believe what you want but only Intel know how long this issue has existed.

There is also the huge elephant in the room that by the time these launch the new X3D chips should be in the channel. I guess it makes sense though if you want to have stability or your CPU has degraded by the time these launch so that the expense is not as great. This makes them not recalling these chips make more sense in a world where the biggest tech tubers are now hated by Intel fans for exposing the truth.

AVX-512 is allegedly disabled due to ensure compatibility with Gracemont E-cores. The underlying Raptor Cove architecture supports it. Remove E-cores, and all you have are full-fat P-cores that should now be able to support AVX-512 without compromise - on a core architecture that is faster than Zen 4 nonetheless. There is no downside to it.

Last year when Intel internally likely already knew that 13th gen were failing? maybe.
And it wont have AVX-512 because that's a workstation upsell feature now nor an useless NPU that no one has figured out a good use case yet.

Speculation aside, I disagree. Intel's keen on being on the AI bandwagon like every other company, and adding this NPU feature to the Bartlett core with Copilot+ compliance (40 TOPS minimum) could be a compelling way to sell more LGA1700 chips, and one-up the Ryzen AI 5 and Ryzen AI 7 segments with 8- and 10 P-core models featuring an NPU and full width AVX-512 VNNI support (instead of the backported 256-bit AVX-VNNI that Raptor Lake has). That said, I'm not sure it's a trivial task to add an NPU to a core design, particularly one that may already have been developed for some time to come, but not released to the public - so this part I wouldn't hold my breath, AVX-512 neural instructions could be all we get. Regardless, they will likely want their Core non-Ultra processor line to have a competitive showing to Ryzens, so they can position their Core Ultra product better. This is what the lack of AVX-512 does to Raptor Lake; as you no doubt know, this processor is otherwise substantially faster than the 7950X.

This is at the Intel "performance" profile, which is the baseline 253W one for my KS processor.

 

[mkppo]

AVX-512 is allegedly disabled due to ensure compatibility with Gracemont E-cores. The underlying Raptor Cove architecture supports it. Remove E-cores, and all you have are full-fat P-cores that should now be able to support AVX-512 without compromise - on a core architecture that is faster than Zen 4 nonetheless. There is no downside to it.

Yeah it was disabled for compatibility reasons but AVX512 workloads in previous gens had to run at reduced clockspeeds and since it's never been enabled and consequently tested on 12/13/14 gen consumer CPU's we can't really say there's no compromise to performance in full fat AVX workloads.

This is what the lack of AVX-512 does to Raptor Lake; as you no doubt know, this processor is otherwise substantially faster than the 7950X.

Which Intel processor is substantially faster than 7950X?

Even ignoring degradation, power consumption and all that, in TPU's latest 8500G review the 7950X sits slightly above the 14900K in overall application performance. Most reviews point to the same, they're pretty equal depending on the applications. In games they're not substantially faster either, and i'm talking about the non X3D parts.

 

[Dr. Dro]

Yeah it was disabled for compatibility reasons but AVX512 workloads in previous gens had to run at reduced clockspeeds and since it's never been enabled and consequently tested on 12/13/14 gen consumer CPU's we can't really say there's no compromise to performance in full fat AVX workloads.



Which Intel processor is substantially faster than 7950X?

Even ignoring degradation, power consumption and all that, in TPU's latest 8500G review the 7950X sits slightly above the 14900K in overall application performance. Most reviews point to the same, they're pretty equal depending on the applications. In games they're not substantially faster either, and i'm talking about the non X3D parts.

There is a compromise to performance - AVX-512 generates a supernova's worth of heat, there is no way you are marketing a 6 GHz+ capable processor with AVX-512 enabled on a 10 nm node. But it's a bit obvious which processor I am referring to, mine. Both the 7950X and 13900KS/14900K should outperform even the 8700G with ease in practically every application with high system requirements.

 

[mkppo]

It's not really just his sample though. Most Raptor Lake owners never had any issues. This would be a lot bigger if every CPU was crashing or worse, which would be a given for a lot of 13th gen i7 and above, in particular the 13900K and KS, which based on their age would already have degraded enough to show issues. Naturally that doesn't mean the issue doesn't exist, or that it isn't big, but it isn't the apocalypse you would think reading the comments in the news about the issue.

I know people have mentioned this already, but the issue here is that degradation comes with use. Consumer CPU's rarely get loaded 24/7 and those will have a drastically different degradation curve with time compared to ones used with 24/7 workloads. This is why most of the complains are coming from server hosting farms/game devs etc.

If a consumer CPU sees 6 hours of full load every day, it'll roughly degrade 4 times less quickly. So the server CPU's that are seeing degradation after 6 months would see the same degradation two years later. It hasn't been two years since these have been launched.

There is a compromise to performance - AVX-512 generates a supernova's worth of heat, there is no way you are marketing a 6 GHz+ capable processor with AVX-512 enabled on a 10 nm node. But it's a bit obvious which processor I am referring to, mine. Both the 7950X and 13900KS/14900K should outperform even the 8700G with ease in practically every application with high system requirements.

I was talking about the 8500G review because that's the latest CPU review from TPU, obviously i'm not comparing 8500G to any of the high end CPU's. That's where the 7950X sits above 14900K in overall application performance so I was confused what CPU you were referring to that's substantially faster than the 7950X.

 

[Dr. Dro]

I was talking about the 8500G review because that's the latest CPU review from TPU, obviously i'm not comparing 8500G to any of the high end CPU's. That's where the 7950X sits above 14900K in overall application performance so I was confused what CPU you were referring to that's substantially faster than the 7950X.

The 7950X supports AVX-512, look at the i9-14900KS review for reference, all tasks which Zen 4 outperforms the Raptor Lake lineup leverage it. Otherwise they are about equal, with a slight lead for the Intel chips, generally, low-thread and very multithreaded (that leverages both P- and E-cores effectively) going for the i9 instead.

Intel Core i9-14900KS Review - The Last of its Kind

The Intel Core i9-14900KS reaches impressive clock frequencies up to 6.2 GHz, which yields remarkable performance in applications and gaming. However, this comes at the cost of serious power consumption, especially if you go beyond stock, where we saw over 500 W during testing.

www.techpowerup.com

AVX-512 makes it notably outperform Raptor Lake in advanced emulation and applications such as V-Ray

 

[mkppo]

The 7950X supports AVX-512, look at the i9-14900KS review for reference, all tasks which Zen 4 outperforms the Raptor Lake lineup leverage it. Otherwise they are about equal, with a slight lead for the Intel chips, generally, low-thread and very multithreaded (that leverages both P- and E-cores effectively) going for the i9 instead.

Intel Core i9-14900KS Review - The Last of its Kind

The Intel Core i9-14900KS reaches impressive clock frequencies up to 6.2 GHz, which yields remarkable performance in applications and gaming. However, this comes at the cost of serious power consumption, especially if you go beyond stock, where we saw over 500 W during testing.

www.techpowerup.com

AVX-512 makes it notably outperform Raptor Lake in advanced emulation and applications such as V-Ray

Yeah exactly, they're about equal so I didn't get the 'substantially faster than 7950X' part and tried to clarify what you meant because it wasn't clear

 

[Dr. Dro]

Yeah exactly, they're about equal so I didn't get the 'substantially faster than 7950X' part and tried to clarify what you meant because it wasn't clear

Raptor Cove IPC is superior, the 16 E-cores are quite competitive with 8 Zen 4 cores - and to compound it, it can run at frequencies that Raphael cannot. Yet it performs at half on the new AIDA benchmark (AVX-512 optimized).

 

[Tomorrow]

Speculation aside, I disagree. Intel's keen on being on the AI bandwagon like every other company, and adding this NPU feature to the Bartlett core with Copilot+ compliance (40 TOPS minimum) could be a compelling way to sell more LGA1700 chips, and one-up the Ryzen AI 5 and Ryzen AI 7 segments with 8- and 10 P-core models featuring an NPU and full width AVX-512 VNNI support (instead of the backported 256-bit AVX-VNNI that Raptor Lake has).

Supposedly Bartlett Lake is meant for high frequency trading etc. That market does not care about NPU.

That said, I'm not sure it's a trivial task to add an NPU to a core design, particularly one that may already have been developed for some time to come, but not released to the public - so this part I wouldn't hold my breath,

That's my take. Adding NPU to a monolithic design would require complete restructuring of the core. Im not sure Intel will go this route as NPU craze seems to be affecting mostly laptops. Desktop models have far weaker NPU's (if any at all).

this processor is otherwise substantially faster than the 7950X.
This is at the Intel "performance" profile, which is the baseline 253W one for my KS processor.

View attachment 356729

This is what supporting proper instruction does. A CPU with lower core/thread count running at lower clock and RAM speeds on a much cheaper motherboard is over twice as fast. Intel decided with 12th gen that offering AVX-512 for consumers was costing them in the workstation/server market and thus they disabled it.

I know people have mentioned this already, but the issue here is that degradation comes with use. Consumer CPU's rarely get loaded 24/7 and those will have a drastically different degradation curve with time compared to ones used with 24/7 workloads. This is why most of the complains are coming from server hosting farms/game devs etc.

If a consumer CPU sees 6 hours of full load every day, it'll roughly degrade 4 times less quickly. So the server CPU's that are seeing degradation after 6 months would see the same degradation two years later. It hasn't been two years since these have been launched.

Return rate has increased for consumers too. Maybe these people run home servers or something, i dont know.
Overall it's a bad look if a CPU degrades the most with use. All silicon degrades with use and always has but not to this extent.
Normally the voltage has been the number one reason for degradation.

 

[Sunny and 75]

Maybe Bring Cyrix/UMC back

XD!

In all seriousness though, I wish both of them the best of luck with their new HW launches whether it's ARL or Z5X3D.

This Q4 will be an exciting one, for sure.

 

[bobbybluz]

Intel replaced my 13600KF without even asking for proof of purchase. That CPU was originally owned by Space Lynx then ShrimpBrime who I got it from. It was fantastic until it wasn't; it ran stable and cool at 5.7GHz for a couple of years. I tried it in an identical mobo and had the same problems with it. I stuck a 12600K in the original mobo and either the board had damaged the 13600KF or vice versa. Trying the 12600K in the other board showed no problems. I also RMA'd the first board and the replacement is supposed to be here Tuesday. Intel and ASRock replaced both quickly with no questions or comments.

I have an original 12700K with AVX-512 stashed away. Even the 12600K outperformed it for A/V work and that 13600KF was in a totally different class of performance than either Alder Lake. At least the 12600K is good for 5.3GHz, the 12700K won't overclock at all. My 5.8-5.9GHz 14700K rig is stashed away for the moment until the chaos here subsides as is my 5.6GHz 13700K PC. At the moment I'm using my old Asus Sabertooth X79 with a 1680 V2 in it as my daily driver. 64GB of DDR3 2100 and running at 4.5GHz it's amazing at how well it works considering its age. It may not be as blistering fast as my Z690 & Z790 PC's but it's easy to cool and always works.

 

[Minus Infinity]

No Intel are saying all 65W and higher cpu's affected including non K. Bartlett Lake and Sapphire Rapids affected too. So it's not the E-cores, it's apparently the ringbus.

 

[Dr. Dro]

No Intel are saying all 65W and higher cpu's affected including non K. Bartlett Lake and Sapphire Rapids affected too. So it's not the E-cores, it's apparently the ringbus.

Has Bartlett released at all across any segment? It's a rumored future processor, that can't be right.

Sapphire Rapids I don't know, did not hear anything about it

 

[remixedcat]

Confirmed issue on laptops

 

[JustBenching]

Has Bartlett released at all across any segment? It's a rumored future processor, that can't be right.

But MLID says bartlett is affected and no microcode can ever fix the issue. Again, don't want to post any links and give clicks, you can find it on his channel in youtube.

Confirmed issue on laptops
View attachment 356748

View attachment 356749

View attachment 356750

Not every problem users experience has to be linked to this. My 6900hs laptop just cold boots on it's own. If it was an intel CPU i'd think it's linked to this, but it's not. Let's all take a deep breath.

 

[stimpy88]

Confirmed issue on laptops
View attachment 356748

View attachment 356749

View attachment 356750

Wow that thing sucks the volts like a CPU from 20 years ago!

 

[Minus Infinity]

Has Bartlett released at all across any segment? It's a rumored future processor, that can't be right.

Sapphire Rapids I don't know, did not hear anything about it

Bartlett Lake is Raptor Lake minus E-cores, same ring bus issues, Sapphire Rapids runs at lower clocks, so symptoms might be milder, less voltage being used.

 

[Dr. Dro]

But MLID says bartlett is affected and no microcode can ever fix the issue. Again, don't want to post any links and give clicks, you can find it on his channel in youtube.


Not every problem users experience has to be linked to this. My 6900hs laptop just cold boots on it's own. If it was an intel CPU i'd think it's linked to this, but it's not. Let's all take a deep breath.

I thought people already knew better than just trust what leakers and rumor mill says... thought wrong I guess.

Bartlett Lake is Raptor Lake minus E-cores, same ring bus issues, Sapphire Rapids runs at lower clocks, so symptoms might be milder, less voltage being used.

Regardless - it's not launched yet and Intel can fix it in hardware before release. Win-win.

 

[JustBenching]

I thought people already knew better than just trust what leakers and rumor mill says... thought wrong I guess.

Just because path of titans (the game made by alderon, the "100% failure rate" guys) crashes even on consoles (you can find it all over on their reddit page) doesn't mean their 100% claims aren't reliable. Come on now

 

[remixedcat]

Wow that thing sucks the volts like a CPU from 20 years ago!

Yes others said that too lol

 

[usiname]

Confirmed issue on laptops

Lol, this laptop is preparing itself for LN2 run