News Raptor Lake instability hits streamer Asmongold — Black Myth: Wukong shader compilation fails on stream

[dalek1234]

This statement is FALSE:

"Intel has already patched Raptor Lake's instability issues through a new microcode update, which not only rectifies CPU instability but also addresses CPU degradation issues caused by excessively high voltages."

The patch only reduces the instability slows down the degradation. The instability and degradation continues to be a thing, just to a lesser extend. What this basically means that you'll see less frequent crashes and your CPU will die a bit later than it would have without the update. It's a design and/or manufacturing issue that is permanently baked into the CPU.

 

[Joseph_138]

Intel went too far, too fast after 11th gen, and it bit them in the butt. They should have lingered for a bit on LGA 1200, while they did more work on the replacement CPU's before moving forward.

 

[rluker5]

You just can't keep increasing the voltage. IF the chip was degraded for excessive voltage.. you're just making things worse.

It seems you aren't aware of the scale of motherboard VRM undervolting. The motherboard manufacturers have been using a hardware/settings combo that provides excessively high low use volts and vdroops to inadequate volts under load. The too low volts make instability while the too high volts promote degradation. To fix the instability you just have to get the volts stable and to stop the degradation you keep the volts out of the danger zone. Normal overclocking basics. But these are a lost art to the vast majority of techtubers and tech sites out there.

I'm talking about increasing the vcore to less than 1.3v for 5.5GHz all core loads with single core 5.8 GHz boost being less than 1.35v on my 13900kf.

1.35v is less than 1.55v. 1.4v should be fine.
Over 1.55v has been known to degrade chips for the last 10 years while 1.35v has been known to be safe for that time. Information that has once again been lost to the ages for the majority of tech media.

But there is a source for quantitative information on system voltages under use. Actually Hardcore Overclocking's youtube has had a bunch of videos on it lately. Even using an oscilloscope to catch all of the transients. Everyone else (other than Jufes) is making qualitative statements that sound like the blind men and the elephant. Somebody looking at the voltage numbers isn't blind. If you are curious you could check out some of Buildzoid's recent videos. Lots of objectively true information there. It really isn't hard to understand the issues if you stop conflating too low voltages (instability) with too high of voltages (leads to degradation). High is different than low.

 

[NinoPino]

So it's a matter of interpretation.

No.

Call it whatever you want and I agree with the "if it's already borked, you better replace it" for the record, but Intel's wording makes it so that anyone who has a "problem free" RaptorLake after this patch (0x129), it should be ok going forward, which is what the article is refering to. At least that is how I interpert the wording.

Here you interpreted correctly, but how can you be sure that at the moment of the upgrade with new microcode if the CPU is really "problem free" ?
It can already be damaged by one month without knowing it, or it can be damaged the day before the upgrade.
And the synthoms (instability) can presents days after by the right workload, or months after because the damage was of minor entity.
In case of light loads on the CPU, the user can also have a stable system even with a degraded CPU, and will never know of it.

 

[NinoPino]

I still say it’s weird that this only happens with nVidia cards in games that use the latest Unreal Engine update that released a few months back when compiling DX12 shaders. These devs like to point fingers and wine and preemptively give notice about the errors, but none have done any troubleshooting at all. Or, at least none that they have posted about. Maybe start by disabling shaders in groups then test to narrow it down to see if there is a specific one or two that cause this issue?

I'm sure that they spent countless hours debugging the problem.
You are talking of Epic and nVidia, for real you think that their developers was not able to correctly troubleshoot the problem ?

Look, at this point (if it’s truly so rampant and widespread) I don’t see why they haven’t done this in an attempt to at least let their games run on these systems without errors or less errors. Or, maybe just give players the chance by adding options to disable specific shaders on their own and test for themselves.

There is no solution a developer can attempt because today can skip the problem in one way for a specific damaged CPU and tomorrow the problem can arise in another point with the same CPU. Not talking of other CPUs damaged in different ways. And for all the problem get worst over time.
When a CPU is damaged, also minimally, it is good only as necklace or paperweight.
Secondary, why spend a lot of time (money) to solve the problem of others ?

This whole thing seems so weird. People are just going along with being okay that something does work right because it’s the other guy’s fault.

Not in this case. There are declarations of nVidia and Intel that prove the issue.

 

[NinoPino]

This statement is FALSE:

"Intel has already patched Raptor Lake's instability issues through a new microcode update, which not only rectifies CPU instability but also addresses CPU degradation issues caused by excessively high voltages."

The patch only reduces the instability slows down the degradation. The instability and degradation continues to be a thing, just to a lesser extend. What this basically means that you'll see less frequent crashes and your CPU will die a bit later than it would have without the update. It's a design and/or manufacturing issue that is permanently baked into the CPU.

I suppose the article referring to two distinct fix. The first was the eTVB bug that cause instability, the second is the over voltage that cause degradation.
Both included into the last microcode.

 

[-Fran-]

Here you interpreted correctly, but how can you be sure that at the moment of the upgrade with new microcode if the CPU is really "problem free" ?
It can already be damaged by one month without knowing it, or it can be damaged the day before the upgrade.
And the synthoms (instability) can presents days after by the right workload, or months after because the damage was of minor entity.
In case of light loads on the CPU, the user can also have a stable system even with a degraded CPU, and will never know of it.

You can't find out easily and that's why it irks me Intel is not issuing a recall. They put the burden of "finding out" on the consumer. That's absolute scum behaviour.

Regards.

 

[cyrusfox]

You can't find out easily and that's why it irks me Intel is not issuing a recall. They put the burden of "finding out" on the consumer. That's absolute scum behaviour.

Regards.

as an owner of a 13900 who uses it mostly for encoding, I am following this closely. I only play quite old games (SC2) and have yet to experience issues (although I have had some IO issues, like extreme slowdown on USB transfers). Always hard to know if the issues are my platform or just windows as if I boot a live USB of linux everything just works...

Really hoping in the coming weeks/month there is a more conclusive way to determine degradation/impact. I am on a H670 board and it has very limited Voltage control. I was personally looking to upgrade to arrowlake, but also interest in bartlett if that is released (12P cores) as an eventual upgrade to this PC.

 

[Eximo]

Alder -> Bartlett is my current hope. Didn't think I would get that lucky. Every other time I have bought an Intel board with the intention to upgrade they have managed to make it incompatible which has kept me off cycle for a long time.

No follow on for LGA1366
Z87 chipset wasn't compatible with Broadwell, only Z97
Z270 chipset was supplanted by a different version of LGA1151
10th i9-11th gen was a non upgrade
I originally intended to get a 14700K, but since they announced Bartlett lake, I'll wait for that. It will be my luck, not compatible with Z690, just you wait.

 

[pf100]

The wording in the article I'm quoting here makes it sound like the microcode update is a fix when Intel never said that it was a fix. To be fair neither did you but you still make it sound like it is. It's only a bandaid to slow but not stop degradation and you should have stated as much in the article.

From the article: "Intel has already patched Raptor Lake's instability issues through a new microcode update, which not only rectifies CPU instability but also addresses CPU degradation issues caused by excessively high voltages."

 

[pf100]

He has it correct. Unless that is a brand new CPU and had the microcode applied to it before much use, it may already have damage that this game exacerbates.

The fix also has to be applied, and there will be people out there who miss it. And they will likely need a CPU replacement, but may not know they can get one.

Already seen a few posts where people have this problem, already took the computer to a shop and were told they had bad memory and other things without having knowledge of the ongoing issue.

Intel never said the microcode update was a fix.

 

[NinoPino]

The wording in the article I'm quoting here makes it sound like the microcode update is a fix when Intel never said that it was a fix. To be fair neither did you but you still make it sound like it is. It's only a bandaid to slow but not stop degradation and you should have stated as much in the article.

From the article: "Intel has already patched Raptor Lake's instability issues through a new microcode update, which not only rectifies CPU instability but also addresses CPU degradation issues caused by excessively high voltages."

I suppose they refers to the fix for the eTVB bug that generate instabilities and was previously fixed with another microcode update. The last microcode include that fix, so it is true that for those that not have applied the 0x125, this microcode improves both stability and fix the voltage problem.