PSA for 13th/14th gen Intel CPU owners

[BugFreak]

Update to the latest BIOS, as it fixes the issue in some way (not fully, but on some brands, it sets slightly different power limits), and wait for the new version with an updated microcode.

On the other hand, this issue has been with us for two years, and somehow, people have been pushing these CPUs to the limits without significant problems (except for overheating). Until earlier this year, barely anyone noticed any problems. This just suggests that the real issue doesn't affect many CPUs or is not visible during work. If not for the noise on the web, I bet 99.5% of users would be happy, not knowing there is any issue.
There is also the fact that during full load, most 13/14 K series CPUs throttle during high load = run at lower (read safe) voltage to keep the low enough temps. So, as it sounds funny, if you have a bad cooler, the CPU runs slower but shouldn't show instability.

That's a good point. I think Intel is quickly becoming the Boeing of the IT world right now, at least according to the media. Every small thing is blown out of proportion and targeted for days of reports. The investment media which I watch/read is really beating them up daily. I hope the damage done isn't too long lasting though.

 

[Woomack]

They already fired 10k employees (I guess it was planned before) and will probably bump prices to cover the losses.

 

[EarthDog]

at least according to the media.

This. I'd hope for more independent thinking at a forum/site such as ours then to get in line with the other lemmings (not saying you are, bug). I think Woomack put a great perspective on things.

I don't find them anything like Boeing. It's not like there have been a constant spate of problems with them (or am I under a rock?).

 

[Flamethrower1972]

I believe in the end everything will be fixed and worked out.

 

[Kenrou]

"Why We Can't Recommend Intel CPUs - Stability Story So Far"

 

[BugFreak]

They already fired 10k employees (I guess it was planned before) and will probably bump prices to cover the losses.

The announced cutbacks should have happen a long time ago. They can't really bump prices too much though because AMD is right there to swipe up market share. Its not like the old days when Intel had a clear advantage over AMD so cutbacks should have happen to reduce cost while keeping the prices and profits in line. Now they have to.

 

[]-[itman]

Update to the latest BIOS, as it fixes the issue in some way (not fully, but on some brands, it sets slightly different power limits), and wait for the new version with an updated microcode.

On the other hand, this issue has been with us for two years, and somehow, people have been pushing these CPUs to the limits without significant problems (except for overheating). Until earlier this year, barely anyone noticed any problems. This just suggests that the real issue doesn't affect many CPUs or is not visible during work. If not for the noise on the web, I bet 99.5% of users would be happy, not knowing there is any issue.
There is also the fact that during full load, most 13/14 K series CPUs throttle during high load = run at lower (read safe) voltage to keep the low enough temps. So, as it sounds funny, if you have a bad cooler, the CPU runs slower but shouldn't show instability.

The problem is that the issue seems to be accelerated degradation, meaning the problem gets worse over time, hence more and more people having issues as time goes on. Additionally, many people who had chips that were effected, didn't realize it until they tried one of the latest UE5 games, because the game engine actually started checking for correct output during decompression and shader compilation. In other words, the CPU was already unstable, they just didn't run any workloads that exposed the instability, or they did but the outputs were never verified for correct results and so they just continued on, not knowing the CPU was putting out wrong outputs.

Now, people who see no issues with their CPU make not have the issue, or it may be that they do have an issue but they don't run workloads that expose the issue and check for correct outputs at the same time. If it is the latter, then they are risking both silent data corruption or issues developing over time, as well as the instability getting worse over time. This is why people were demanding an increased warranty window, which thankfully Intel has agreed to, at least for retail boxed processors.

Lastly, Intel has been changing its story over time about which CPUs are effected and what the problem is, which unfortunately makes people suspicious of future statements and solutions Intel will provide. It's not the end of the world for Intel by any means, but it is a real problem, and one that is going to end in (probably) multiple lawsuits. There are already multiple reports of RMA issues and lack of replacement inventory, even when RMAs are approved. It's going to be a real pain for Intel now, and the near future. The good news, is that Arrow Lake should hopefully not have any of these issues and the faster they can move on to it, the better for Intel. The bad news there is that Arrow Lake is going to be made, by and large, on TSMC's N3B node, so it's going to be costly for Intel until they can transition to their own 18a process for the next gen CPUs.

 

[Kenrou]

Agree with hitman, the way I see it, the damage is already done if you've been using your CPU for a while, the crux is if the problem is large enough to cause issues during your daily running or not. Some will be fine(ish) because they were barely used, others will be fine until you push them (like the case of UE5), others will be crashing non-stop, and the cause is still either Intel or the partners because of the high voltages. Either way, (whether you care for it or not) accelerated degradation will cause your chip to become unstable much sooner than you would expect it to, especially if you continue to push it.

And the added problem: will you be one of the lucky ones that Intel replaces your chip?

 

[EarthDog]

Either way, (whether you care for it or not) accelerated degradation will cause your chip to become unstable much sooner than you would expect it to, especially if you continue to push it.

And the added problem: will you be one of the lucky ones that Intel replaces your chip?

Perhaps... but if it lasts the warranty, I'm not going to ***** about it otherwise.

Only time will tell on the RMAs. I see too much speculation. I'll keep the glass half full and have some faith. Ya'll can crap all over it (understandably).

or it may be that they do have an issue but they don't run workloads that expose the issue and check for correct outputs at the same time.

If a tree falls in the forsest and nobody is there to hear it.........

 

[]-[itman]

Perhaps... but if it lasts the warranty, I'm not going to ***** about it otherwise.

Only time will tell on the RMAs. I see too much speculation. I'll keep the glass half full and have some faith. Ya'll can crap all over it.


If a tree falls in the forsest and nobody is there to hear it.........

Agree on the RMAs. I don't think it will be an issue in the end, but I do think people should expect to be given a hard time in trying to get an RMA as this goes on (already have multiple people online saying they got push back from Intel at first). As long as people just keep pushing and, "don't take no for an answer," I expect Intel will honor their RMA requests and I think bumping the warranty period to 5 years is a good way to compensate for the issue.

As far as the forest comment, it comes down to silent data corruption (ignoring that you said this at least partially in jest just to clarify the issue). If it is an unstable CPU, but no work loads with integrity checks are being done, that means you will most likely have data corruption that will accumulate over time. This used to be something overclockers would talk about when discussing stability; just because you don't see any signs of instability in your day to day usage, doesn't mean it isn't happening behind the scenes. The symptoms aren't obvious but can slowly lead to corrupted files, needing to reinstall programs, Windows features all of a sudden not working, etc.

 

[EarthDog]

I expect Intel will honor their RMA requests and I think bumping the warranty period to 5 years is a good way to compensate for the issue.

I'm sure some people will have to jump through hoops, it's inevitable. Hopefully, it's not any more than usual. People are, right or wrong, going to be RMAing these in droves though. Because the issue does exist, people will now have (rightfully so?) an itchy trigger finger at any hint of instability. I do not envy their RMA department. Oof.

As far as the forest comment, it comes down to silent data corruption (ignoring that you said this at least partially in jest just to clarify the issue). If it is an unstable CPU, but no work loads with integrity checks are being done, that means you will most likely have data corruption that will accumulate over time.

Surely, especially for a 'normal' user (the overwhelming majority) who probably doesn't reinstall their OS (ever!) or nearly as often as, say, overclockers (enthusiasts). Again, Intel is in a pickle, at least in their customer service and RMA departments.

 

[]-[itman]

I'm sure some people will have to jump through hoops, it's inevitable. Hopefully, it's not any more than usual. People are, right or wrong, going to be RMAing these in droves though. Because the issue does exist, people will now have (rightfully so?) an itchy trigger finger at any hint of instability. I do not envy their RMA department. Oof.


Surely, especially for a 'normal' user (the overwhelming majority) who probably doesn't reinstall their OS (ever!) or nearly as often as, say, overclockers (enthusiasts). Again, Intel is in a pickle, at least in their customer service and RMA departments.

Yeah, that will definitely be an unfortunate knock-on effect of all of this. It's been a long time since people (who didn't overclock) ever had to worry about their CPU being the source of instability/errors. Now, any time someone with a RPL CPU has a crash or something weird happens, they will probably blame the CPU, even if the CPU is fine and it's actually a driver issue or bad memory or something.

 

[mackerel]

Intel has been changing its story over time about which CPUs are effected and what the problem is

Have they? As far as I'm aware they have only stated one problem: the voltage thing. What else is there? The oxidation thing is basically noise at this point. I've not followed affected list that closely, and the only "discrepancy" I've seen claimed elsewhere but not verified is 65W vs the latest list. This may be in part due to some parts being rebadged Alder Lake.

 

[]-[itman]

Have they? As far as I'm aware they have only stated one problem: the voltage thing. What else is there? The oxidation thing is basically noise at this point. I've not followed affected list that closely, and the only "discrepancy" I've seen claimed elsewhere but not verified is 65W vs the latest list. This may be in part due to some parts being rebadged Alder Lake.

They initially said that only unlocked i7 and i9 processors were affected, then later that expanded to basically all 65+W RPL processors. They also initially tried to say that it was the motherboard makers’ fault for not following Intel guidelines, though this was a whole mess of messaging because they kept changing what the official guidelines were and publicly only insinuated it was a motherboard issue.

Lastly, the oxidation issue they initially said was found in 2023 and was fixed in 2023 and was not a cause of instability in RPL processors. They later said it was a cause of instability, but only a small number were affected. That again got updated to Intel discovering the issue in late 2022 and not fully addressing it until early 2024, and that there may still be CPUs on the shelf that were manufactured with this defect.

 

[mackerel]

Without checking exactly what they actually said, as opposed to what people think they said, the timeline seems consistent. As more info was obtained they updated. We have to split this into two time periods: before and after root cause. Problem was reported. First point at controlling it were non-aggressive mobo settings. Troubleshooting 101? Set safe defaults and work from there. Without having a confirmed cause at the time, they could only make statements on what was reported to have problems. As root cause was found and understanding of it improved, they refined their statements. They can't tell you what they don't know themselves.

It is pretty much a worst case problem to have. A longer term gradual degradation isn't something you can easily test, and I think this contributed to the timescales involved from when it was first reported to announcing root cause. They look at possible reasons what is wrong. Even if they think they found the voltage was incorrectly set, to confirm it they likely would have to replicate it to be sure it matches the symptoms. It would be worse for them to admit to a problem that isn't there.

 

[]-[itman]

Without checking exactly what they actually said, as opposed to what people think they said, the timeline seems consistent. As more info was obtained they updated. We have to split this into two time periods: before and after root cause. Problem was reported. First point at controlling it were non-aggressive mobo settings. Troubleshooting 101? Set safe defaults and work from there. Without having a confirmed cause at the time, they could only make statements on what was reported to have problems. As root cause was found and understanding of it improved, they refined their statements. They can't tell you what they don't know themselves.

It is pretty much a worst case problem to have. A longer term gradual degradation isn't something you can easily test, and I think this contributed to the timescales involved from when it was first reported to announcing root cause. They look at possible reasons what is wrong. Even if they think they found the voltage was incorrectly set, to confirm it they likely would have to replicate it to be sure it matches the symptoms. It would be worse for them to admit to a problem that isn't there.

I know he tends to be over dramatic about things, but the GN videos have documented all of this stuff really well.

I agree that Intel's understanding of the problem has evolved with time, but from a PR perspective, it erodes trust when you first say that only certain products are vulnerable, but then later have to expand the product list significantly, even if it was just a lack of Intel themselves knowing the full scope of the problem. For the oxidation issue, I can't think of a charitable explanation for their story changing there. This whole thing has been as much of a bungling of the PR as it has been the actual problem.

Even with their own partners, they've had a severe lack of communication on the issue to the point that motherboard makers were actually asking tech reviewers what settings they should use after Intel said that motherboards should be set to follow Intel specs but then told the motherboard makers that they didn't have an official spec for them and left them in the dark for months. Long term, I don't think this hurts Intel that much as long as they honor their extended warranties (which I fully expect them to do), but now and the near future is going to be a bit rough.

 

[Kenrou]

Well, for good or for bad it's likely to get overshadowed by AMD "tweaking" benchs (again) to show the "new" CPU lineup to be as fast as certain Intel (which they certainly aren't)...

 

[]-[itman]

Well, for good or for bad it's likely to get overshadowed by AMD "tweaking" benchs (again) to show the "new" CPU lineup to be as fast as certain Intel (which they certainly aren't)...

Are you talking about the XT chips that just launched or the upcoming Zen 5 launch chips?

 

[Kenrou]

The XT, which according to reviews are literally the same chips as the X with just another 100mhz on top.

 

[]-[itman]

The XT, which according to reviews are literally the same chips as the X with just another 100mhz on top.

You get 4 more cores for the 5900XT vs. the 5900X as well, but yeah, AMD basically tested under a GPU bottleneck to make their claim, which people spotted right away. The reviews are already out for it, and as expected, it can't keep up with RPL in gaming or many other work loads. I don't think most people really care, though, considering it is a refresh of CPUs from 2 generations ago.