Unless Intel does something huge (which I doubt since AMD is an astonishing 3x better perf/watt with X3D right now), I'll be doing a 9800X3D build next. My 5800X has been great for almost 4 years.

RogueSix wrote on Aug 4, 2024, 20:43:

Thanks but errr... ahem... so the source is a shi... umm... I mean, an influencer instead of Intel directly?

I started watching that video and after a grand mal seizure i noticed he had less than 32,000 subscribers. But the coup de grâce, he bills himself as a gamer and tech focused streamer.Imagine visiting a dentist and the credentials on the wall say..."Flosses regularly and brushes twice a day." With no mention of the school he attended or degree awarded.

Kxmode wrote on Aug 4, 2024, 19:37:

RogueSix wrote on Aug 4, 2024, 09:38:
The nVidia installer thing is extremely interesting for me personally. I have zero issues with stability in games (including shader compilation) or benchmarks or stress tests. But one issue I have had that started early this year is that my nVidia drivers setup would randomly fail sometimes with a "nVidia installer failed" error message.

I did NOT (never) get the "out of memory" error that some people have been reporting but always just the non-descript "nVidia installer failed" error message. I did a lot of troubleshooting/experimenting but it was totally random. IIRC, it started in early 2024 and it's been happening sporadically ever since. The latest drivers install (560.70) went through without issue but there were a few really annoying times previously when it took a dozen attempts or so before the install would work.

Would you happen to have a link where Intel suggested to install the nVidia drivers five times in a row and where they explain that errors are/were related to the eTVB bug? I could not find anything about that after a quick Google search.

How to check your CPU. Apologies, he doesn't explicitly say "Intel recommends" but he does note that the NVIDIA driver installation uses the same Oodle decompression (that can trigger the the BSOD as I've experienced).

Thanks but errr... ahem... so the source is a shi... umm... I mean, an influencer instead of Intel directly?

In that case, I'm not convinced if my very sporadic issues have anything to do with the CPU. Besides, I always do a manual install of the nVidia drivers, i.e. I manually extract the drivers package to my desktop with 7-zip with zero errors 100% of the time.

Then I manually delete all the superfluous stuff/folders/bloat and double click the setup.exe. That is when I always select a custom install, deselect GFE and check the 'clean install' box.

I have received the non-descript "nVidia installer failed" error message at completely random points during the setup routine. Sometimes during the compatibility check right at the start, sometimes when the install routine was removing the old drivers (clean install) and sometimes when it was installing the new drivers. There was absolutely no pattern to it.

I somehow doubt that any decompression is taking place at every single step here where the install has been failing for me so I'm once again more inclined to believe that my issues might have been software related somehow. As I said, the CPU is otherwise 100% stable so I guess I might be off the hook for now but that won't change the fact that I will switch to Zen 5 + X870E ASAP (pleeeeaaaase hurry up AMD... I can't live with this trash anymore! ).

Prez wrote on Aug 4, 2024, 16:18:
I almost missed your reply BoP because it was at the end of a different reply. I wasn't very clear.; I've been around the block enough to know that the easiest solution isn't usually the best, and the best solution is rarely the easiest. What I meant to say was that I wanted to implement the best possible solution that I can manage. It probably will end up being a trade-off between effective enough while being easy enough. Regardless, I won't be blindly buying an Intel product ever again. Hopefully I can avoid ever buying anything from them ever again, but that depends on AMD and/or whatever alternatives exist at the time being functionally viable.

It's all good.

Good fixes are hard, which I am sure you already well know given what you've said about your past employment around nuclear power.

Honestly, I'm rooting for the M architecture from Apple to lead the way in to the next stage of computing. I have nothing but respect for what Apple has managed to pull off with ARM. I have said it in the past but I will say it again. Microarchitecture like ARM (but maybe not ARM itself) is the future of desktop computing. It's already heavily used in enterprise cloud deployments and mobile products. Low power but performant for everyone but the most power hungry user. Especially with the looming energy crises, low power is going to probably be a forced option.

RogueSix wrote on Aug 4, 2024, 09:38:
The nVidia installer thing is extremely interesting for me personally. I have zero issues with stability in games (including shader compilation) or benchmarks or stress tests. But one issue I have had that started early this year is that my nVidia drivers setup would randomly fail sometimes with a "nVidia installer failed" error message.

I did NOT (never) get the "out of memory" error that some people have been reporting but always just the non-descript "nVidia installer failed" error message. I did a lot of troubleshooting/experimenting but it was totally random. IIRC, it started in early 2024 and it's been happening sporadically ever since. The latest drivers install (560.70) went through without issue but there were a few really annoying times previously when it took a dozen attempts or so before the install would work.

Would you happen to have a link where Intel suggested to install the nVidia drivers five times in a row and where they explain that errors are/were related to the eTVB bug? I could not find anything about that after a quick Google search.

How to check your CPU. Apologies, he doesn't explicitly say "Intel recommends" but he does note that the NVIDIA driver installation uses the same Oodle decompression (that can trigger the the BSOD as I've experienced).

This comment was edited on Aug 4, 2024, 19:55.

I almost missed your reply BoP because it was at the end of a different reply. I wasn't very clear.; I've been around the block enough to know that the easiest solution isn't usually the best, and the best solution is rarely the easiest. What I meant to say was that I wanted to implement the best possible solution that I can manage. It probably will end up being a trade-off between effective enough while being easy enough. Regardless, I won't be blindly buying an Intel product ever again. Hopefully I can avoid ever buying anything from them ever again, but that depends on AMD and/or whatever alternatives exist at the time being functionally viable.

RedEye9 wrote on Aug 4, 2024, 11:08:

I have no stability issues except for the "nVidia installer failed" test.

So what you're saying is your CPU didn't pass Intel's recommended stability check.
That right there should tell you all you need to know.
Let us know if they approve your RMA?

Where did they recommend this? Could anyone provide a link please?

As I said, I'm probably just going to eBay (with corresponding disclaimers since I'm an honest seller) my piece of shit degraded CPU as soon as Zen 5 + X870E is on shelves.

If you found Steve's video too long, here's the CliffsNotes version from Paul's Hardware. The sarcasm is strong with this one...

https://www.youtube.com/watch?v=wnNjjAit-eY

I have no stability issues except for the "nVidia installer failed" test.

So what you're saying is your CPU didn't pass Intel's recommended stability check.
That right there should tell you all you need to know.
Let us know if they approve your RMA?

The Flying Penguin wrote on Aug 4, 2024, 09:57:
The voltage issue and the via corrosion issues are two separate things (as far as we know). I understand that the former will take time to resolve (write and test the new microcode), but c'mon Intel. You've KNOWN about the corrosion issue for AT LEAST a year (by some reports longer).

There is zero excuse not to release a list of dates and/or lot numbers so OEMs and end users can check to see if they have affected CPUs.

Exactly!

I might have have an oxidation affected CPU since I bought mine in late December 2022 (built the new rig literally on New Year's day, i.e. 01/01/2023). As I said, no (stability) issues except for that "nVidia installer failed" stuff but Intel should definitely at the very least offer a recall for those CPUs that were affected by oxidation.

There is also a good chance that Intel is up to dirty tricks with regard to the microcode fix. They seem to want to wait until after the Zen 5 NDA so that most reviews will compare Zen 5 with the old (now proven unsafe/unstable) Raptor Lake/Raptor Lake Refresh results.

It will be interesting to see how much performance that microcode fix is going to shave off and I'm wondering if AMD delayed Zen 5 from the original July 31st release not for the purported quality issues but simply to cross Intel's devious plans. It's like they are staring each other down right now... who is gonna make the first move? Microcode or Zen 5 release?

Depending on timing, I really hope that most review sites will not fall for Intel's shit. They should only compare Zen 5 to Zen 4 in their reviews at first and then update their reviews with updated Intel results *after* the microcode fix has been released.

The voltage issue and the via corrosion issues are two separate things (as far as we know). I understand that the former will take time to resolve (write and test the new microcode), but c'mon Intel. You've KNOWN about the corrosion issue for AT LEAST a year (by some reports longer).

There is zero excuse not to release a list of dates and/or lot numbers so OEMs and end users can check to see if they have affected CPUs.

Kxmode wrote on Aug 3, 2024, 21:59:

RogueSix wrote on Aug 3, 2024, 20:29:

Kxmode wrote on Aug 3, 2024, 19:02:
I'm in the middle of this nonsense, focusing on finding fixes. But I'm also distraught with Intel. I'll likely go with AMD the next time I upgrade. The good news is that all the fixes I've seen have done a great job, so I'm hopeful that the final patch will 100% fix the problem, and I can finally run my CPU at the default P-Core multipliers.

Oooohhh... so you are personally affected? Didn't you actually upgrade just a couple of months ago?

Yup. I finished building my brand new 14900K PC, and about a month later, all this noise started to appear. You were actually the one that encouraged me to upgrade. 🙂

Well, I wholeheartedly (seriously!) apologize about that one in hindsight . I couldn't have known at that time. When I upgraded in January 2023, I was kind of torn between AMD and Intel and I could bite chunks outta my ass in retrospect for going with Intel, of course.

Oh well, I will switch over to AMD as soon as the X870E boards are available. I might not even wait for X3D but just go with a 9700X at first and replace it with a 9800X3D later on. I will eBay my 13900K with huge disclaimers to keep it honest and then I'm done with Intel for the foreseeable future. They would have to pull off some serious miracles before I would ever consider them again.

It shouldn't be degradation but more about the motherboard running the CPU outside of Intel's specs. Intel recommended one way to test for degradation was to run the Nvidia driver installation around five times after installing the BIOS update with the 0x125 eTVB microcode update (that was BIOS version 1402 on my ASUS ROG Dark Hero mobo). NVIDIA was able to install the latest drivers five times without issue. Furthermore, I understand that Intel produced my CPU after they fixed the Via Oxidation issues sometime in early 2023.

So, my one hope is that the upcoming final patch, expected sometime in mid-August, will finally put the CPU in a state where I can reset the P-core multiplier back to its default values and game with zero issues.

Well, if you have to lower the multiplier then you are forced to run the CPU below specs. That is 100% Intel's fault and not the board makers. Also, degradation is not just related to the oxidation issue but, as Intel admitted, affects all desktop CPUs that suffered from those elevated voltages. The degradation is a multi-level/multi-source issue.
It is my understanding that these voltages are something "CPU-internal" which is why a microcode fix is needed. If it would be as simple as limiting voltages in BIOS then Intel surely would have suggested as much.

The nVidia installer thing is extremely interesting for me personally. I have zero issues with stability in games (including shader compilation) or benchmarks or stress tests. But one issue I have had that started early this year is that my nVidia drivers setup would randomly fail sometimes with a "nVidia installer failed" error message.

I did NOT (never) get the "out of memory" error that some people have been reporting but always just the non-descript "nVidia installer failed" error message. I did a lot of troubleshooting/experimenting but it was totally random. IIRC, it started in early 2024 and it's been happening sporadically ever since. The latest drivers install (560.70) went through without issue but there were a few really annoying times previously when it took a dozen attempts or so before the install would work.

Would you happen to have a link where Intel suggested to install the nVidia drivers five times in a row and where they explain that errors are/were related to the eTVB bug? I could not find anything about that after a quick Google search.

To be clear, I'm not doubting what you are saying but I'm genuinely interested in reading up on the technical details because that is the one and only issue I have had with my system in recent months. Everything else is rock stable at PL1 = PL2 = 253W ... I only forbid ASUS to set PL2 to 4096W but otherwise my 13900K is running at its full potential.
However, rock stable or not... this CPU is flying out the window as soon as Zen 5 + X870E are available. Can't be bothered with the bad feeling of running a degraded/degrading CPU anymore. Yuck@Intel.

Prez wrote on Aug 3, 2024, 21:09:
[... can't shake the feeling that I have about Gamers Nexus. It feels identical to the feeling that I have with Larian (makers of Baldur's Gate 3). There was a time when Blizzard was every bit as revered as Larian is right now. There was a time when Linus Tech Tips was almost beyond reproach. I look at companies like Larian and Gamers Nexus and I wonder how long until those companies become the newest companies that we point to and say "they are so hated because they are shit. They are nothing like Company 'X', who cares about quality and cares about the customer." Then further down the line, when Company 'X' falls from grace and becomes hated but here is Company 'Y' to save the day. And so on. Holy shit I sound so cynical but this cycle just keeps happening. I don't see companies that are champions for consumers and doing the right thing; I see companies that haven't quite grown large enough yet to become just as bad or worse as the companies they once were compared favorably to. I don't want to be that jaded and cynical. But unfortunately I haven't been wrong yet.

This cycle is inevitable. A well respected organisation will screw up eventually unless they fade into obscurity. A high profile organisation will always have some detractors. So if a well respected, high profile organisation doesn't fade away, eventually their detractors will have something to crow about and shift the narrative for the 95% of people who don't have particularly strong opinions one way or another.

Nobody can avoid screwups forever. The only question is how big the fall; if it comes with a sense of dishonesty or betrayal the fall can be pretty damn big. LTT generally looked incompetent so they took a pretty big hit but seem to have recovered fine (which imo is exactly what they deserved). Blizzard looked like they betrayed their player-base and it's hard for someone's reputation to recover from that.

Simon Says wrote on Aug 3, 2024, 19:18:
My guess is they've been pushed very hard by management to deliver as much performance as possible as soon as possible so they cut corners everywhere while pushing everything to the redline hoping everything would work out.

I mentioned the P4 upthread and I see a lot of similarities between that era and this one. In both instances, Intel has found itself on the back foot after being simply outperformed by AMD. Also in both instances, it has tried to keep itself afloat by juicing lame duck products until it can flounder around and find something better. Today, though, I think Intel is in a worse position. They have bet the farm on their attempt at a big.LITTLE design (what they call "Performance" and "Efficiency" cores) and, like I also mentioned previously, their track record with those designs is abysmal. They were also slow to get away from monolithic core design. Something AMD had already jettisoned when it went to chiplet design with the 1st gen Ryzen back in 2017.

I think it's a one-two punch that has Intel staggering. The first is being outfoxed and outperformed by AMD. The second being that they shot themselves in both feet and the dick when they insisted on using a lithography method that everyone told them would be a failure, both internally and externally. At the time, Intel pulled an IBM with "we know better. We'll MAKE it work". Of course it didn't. So now they're stuck with a white elephant lithography and a microarchitecture that can only be competitive if it is juiced to insane levels. They're not alone in this, either, AMD is also now juicing their CPUs at the top end to insane levels because I don't think they have a viable successor to Ryzen yet. So they, too, are somewhat stuck on a microarchitecture that is running out of legs, so to speak.

I will cautiously say that I still have friends on the inside and they are frustrated. Not with the public but with the stagnation.

/---\

Prez wrote on Aug 3, 2024, 21:09:
I'm waiting to see what will end up being the best fix (i.e. the one that works best while being the easiest to implement). In the meantime, I can't shake the feeling that I have about Gamers Nexus. It feels identical to the feeling that I have with Larian (makers of Baldur's Gate 3). There was a time when Blizzard was every bit as revered as Larian is right now. There was a time when Linus Tech Tips was almost beyond reproach. I look at companies like Larian and Gamers Nexus and I wonder how long until those companies become the newest companies that we point to and say "they are so hated because they are shit. They are nothing like Company 'X', who cares about quality and cares about the customer." Then further down the line, when Company 'X' falls from grace and becomes hated but here is Company 'Y' to save the day. And so on. Holy shit I sound so cynical but this cycle just keeps happening. I don't see companies that are champions for consumers and doing the right thing; I see companies that haven't quite grown large enough yet to become just as bad or worse as the companies they once were compared favorably to. I don't want to be that jaded and cynical. But unfortunately I haven't been wrong yet.

The best fix is typically not the one that is the easiest to implement and the easiest to implement is typically not the best fix. I do not think there is going to be a "easy and best" fix forthcoming. I think the best that can be hoped for is "This will ease the issue and require little intervention on the part of the average user". It'll be a compromise that will be suboptimal.

I do not think you are cynical or jaded. I think you are acutely aware of the boom and bust cycles of hypercapitalism and have lived through many, many examples of that cycle in your lifetime. Therefore it is easy for you to spot the trends. History doesn't repeat itself but it certainly does rhyme. Every industry "darling" gets bloated, loses sight of its core that got it to where it was, and then eventually fizzles out. This is especially true when shareholders become involved because a company can't say "We're cutting everything back to the core and will be doing what we do best". Shareholders would have a bloody fit because it means they couldn't have their infinite growth and infinite ROI. They don't care if a company dies, they just want to extract as much value as possible before it does. It's rare but you do on occasion find companies that aren't afflicted with the parasitic shareholder. They tend to remain medium sized but just keep knocking it out of the park with their products consistently.

I think Aristotle would probably look at today's business and repeat his well quoted line "“The greatest crimes are not those committed for the sake of necessity but those committed for the sake of superfluity. One does not become a tyrant to avoid the cold".

RogueSix wrote on Aug 3, 2024, 20:29:

Kxmode wrote on Aug 3, 2024, 19:02:
I'm in the middle of this nonsense, focusing on finding fixes. But I'm also distraught with Intel. I'll likely go with AMD the next time I upgrade. The good news is that all the fixes I've seen have done a great job, so I'm hopeful that the final patch will 100% fix the problem, and I can finally run my CPU at the default P-Core multipliers.

Oooohhh... so you are personally affected? Didn't you actually upgrade just a couple of months ago?

Yup. I finished building my brand new 14900K PC, and about a month later, all this noise started to appear. You were actually the one that encouraged me to upgrade. 🙂

RogueSix wrote on Aug 3, 2024, 20:29:
Do you have any hardcore use cases that put your CPU under extraordinary stress?

Before I set the PL1 and PL2 limits, adjusted the amps, and lowered the p-core multipliers in Intel's Extreme Tuning Utility, games were crashing. At least one of these crashes resulted in a BSOD: Cyan's Riven remake, related to the RAD Tools' Oodle Decompression issue. Currently, things are stable, but I still need to keep the p-core multipliers reduced by 4-7; otherwise, games like Valhalla suddenly crash to the desktop.

Strangely, I hardly ever remember seeing any issues with benchmarking apps. It's been mostly games, especially those running on Unreal Engine 5, where I've encountered problems.

RogueSix wrote on Aug 3, 2024, 20:29:
I would take your example with an almost brand-new CPU as further proof that there is more afoot than Intel wants us to believe. There is no way your CPU could have degraded that much in such a short amount of time. I'm not at 100% yet but I'm betting 99.6% that this is simply a hardware defect and the microcode fix is a bandaid to merely mitigate the issue on the surface.

It shouldn't be degradation but more about the motherboard running the CPU outside of Intel's specs. Intel recommended one way to test for degradation was to run the Nvidia driver installation around five times after installing the BIOS update with the 0x125 eTVB microcode update (that was BIOS version 1402 on my ASUS ROG Dark Hero mobo). NVIDIA was able to install the latest drivers five times without issue. Furthermore, I understand that Intel produced my CPU after they fixed the Via Oxidation issues sometime in early 2023.

So, my one hope is that the upcoming final patch, expected sometime in mid-August, will finally put the CPU in a state where I can reset the P-core multiplier back to its default values and game with zero issues.

RedEye9 wrote on Aug 3, 2024, 12:08:
In the before-fore times when Intel knowingly released a faulty CPU they could be shamed into doing the right thing and recall it*, but now consumers enjoy getting bent over without the courtesy of a reach around and we are stuck with this nonsense.


* https://en.wikipedia.org/wiki/Pentium_FDIV_bug
In December 1994, Intel recalled the defective processors in what was the first full recall of a computer chip.


Reading the Discovery and Response section sounds precisely like what's going on as it unfolds in real-time. We're current at the "At this point, Intel acknowledged the floating-point flaw, but claimed that it was not serious and would not affect most users. Intel offered to replace processors to users who could prove that they were affected."

I'm waiting to see what will end up being the best fix (i.e. the one that works best while being the easiest to implement). In the meantime, I can't shake the feeling that I have about Gamers Nexus. It feels identical to the feeling that I have with Larian (makers of Baldur's Gate 3). There was a time when Blizzard was every bit as revered as Larian is right now. There was a time when Linus Tech Tips was almost beyond reproach. I look at companies like Larian and Gamers Nexus and I wonder how long until those companies become the newest companies that we point to and say "they are so hated because they are shit. They are nothing like Company 'X', who cares about quality and cares about the customer." Then further down the line, when Company 'X' falls from grace and becomes hated but here is Company 'Y' to save the day. And so on. Holy shit I sound so cynical but this cycle just keeps happening. I don't see companies that are champions for consumers and doing the right thing; I see companies that haven't quite grown large enough yet to become just as bad or worse as the companies they once were compared favorably to. I don't want to be that jaded and cynical. But unfortunately I haven't been wrong yet.

Kxmode wrote on Aug 3, 2024, 19:02:
... until the final patch...

You seem to be quite the optimist . No one knows for sure (possibly not even Intel) if the microcode fix for the elevated voltages is the "final patch". We will have to wait and see about that. Intel themselves have admitted that any degradation is irreversible so people who have owned RPL CPUs since the beginning, or for any substantial amount of time, are stuck with a degraded CPU in any case.

Best case is indeed that this is the "final patch" so people may be stuck with a degraded CPU but at least there will be no further accelerated degradation on top of the (very slow) natural degradation. However, I would not rule out the possibility that the "patch" will only mitigate the degradation but not stop it.

Due to the nature of the issue, only time will tell. It would actually be a pity if the microcode fix would solve the issue for good. It was just starting to get fun!

Kxmode wrote on Aug 3, 2024, 19:02:
I'm in the middle of this nonsense, focusing on finding fixes. But I'm also distraught with Intel. I'll likely go with AMD the next time I upgrade. The good news is that all the fixes I've seen have done a great job, so I'm hopeful that the final patch will 100% fix the problem, and I can finally run my CPU at the default P-Core multipliers.

Oooohhh... so you are personally affected? Didn't you actually upgrade just a couple of months ago? Do you have any hardcore use cases that put your CPU under extraordinary stress?

I would take your example with an almost brand-new CPU as further proof that there is more afoot than Intel wants us to believe. There is no way your CPU could have degraded that much in such a short amount of time. I'm not at 100% yet but I'm betting 99.6% that this is simply a hardware defect and the microcode fix is a bandaid to merely mitigate the issue on the surface.

Kxmode wrote on Aug 3, 2024, 19:02:
Conversely, this video by Robeytech briefly explains the issue and how to immediately remedy the situation until the final patch comes out in a week or two.

This is just the tentative stopgap which is not 100% effective since it doesn't stop voltage spikes.
If you really want to play it safe you must prevent the CPU from spiking voltage.
The easiest way is to limit the maximum multiplier to something significantly lower than default, say if default is 57, drop down to 53, if default is 53, drop down to 49, and so on.

There are more involved ways that lets you retain performance, you can look them out, they involve AC/DC loadline curve tweaks, there are a few youtube videos on the subject.

Burrito of Peace wrote on Aug 3, 2024, 13:10:
I have to wonder...how the actual fuck did that microcode make it out of validation testing?

It's not like Intel "engineers" didn't openly boast about rushing through the development process at record speed to "beat AMD". /s

My guess is they've been pushed very hard by management to deliver as much performance as possible as soon as possible so they cut corners everywhere while pushing everything to the redline hoping everything would work out.

Guess that didn't turn out so well...

Earlyworm wrote on Aug 3, 2024, 14:57:
I really want to watch the Gamer Nexus videos, he is well spoken, with interesting analysis and subject selection... but man, can he drone on and on.. I'm not watching 47 minutes of tech drama.

You'll likely find this video better as it features Gamer Nexus and Wendell from Level1 Techs having a more in-depth discussion about the issue. It's a bit more lively.