News Intel finds root cause of CPU crashing and instability errors, prepares new and final microcode update

[Admin]

There are four circumstances that can cause Vmin Shift Instability Root Cause, according to Intel.

Intel finds root cause of CPU crashing and instability errors, prepares new and final microcode update : Read more

 

[bit_user]

The drip-feed of these mitigations doesn't inspire a ton of confidence. How do we know they aren't just pacing them out until Arrow Lake launches, at which point they'll release the final mitigation that actually has a significant performance impact.

Intel's internal tests show that the 0x12B update does not noticeably affect performance. Benchmarks and gaming tests, including popular titles like Cyberpunk 2077 and Shadow of the Tomb Raider, showed results within normal expected variations when compared to the earlier 0x125 update.

Yeah, relative to 0x125 - and that includes running at stock power limits and Tau, which is not what a lot of gaming boards defaulted to, or how most reviewers benchmarked it.

I'd love to see how much performance the i9-14900K lost, between those original reviews and this latest update.

 

[Marlin1975]

Hahaha So this time its going to be fixed.... trust intel... haha

I have to agree. This just seems like they are delaying recalls and hope people will forget.

 

[ThisIsMe]

In a world of infinite possibilities, that certainly is a possibility. No need to stop there you two. Keep the imagination flowing. lol

 

[setx]

Look, it's 14+++ again!

 

[rluker5]

I don't know if messing with how load line calibration works is a good thing or not. That is what this microcode update is doing. And it is doing it because the motherboard manufacturers are setting the LLC to have too much vdroop by default.

When you apply a lot of amps to your CPU and the motherboard has set the LLC to a large amount of vdroop the volts drop a lot. But they have to stay high enough to be stable. So the time when the CPU is not running a lot of amps the voltages will be much higher.

If you lower the vdroop you can set your load voltages the same and your low load voltages will be much lower.

What the motherboard manufacturers have done, in an attempt to bend the rules to apply a factory undervolt, is responsible for nearly all of the degradation issues.

 

[TheHerald]

The drip-feed of these mitigations doesn't inspire a ton of confidence. How do we know they aren't just pacing them out until Arrow Lake launches, at which point they'll release the final mitigation that actually has a significant performance impact.


Yeah, relative to 0x125 - and that includes running at stock power limits and Tau, which is not what a lot of gaming boards defaulted to, or how most reviewers benchmarked it.

I'd love to see how much performance the i9-14900K lost, between those original reviews and this latest update.

Why would it have significant performance impact? Even if it's let's say restricted to the 12900k or ks clockspeeds (which supposedly have no such issue) performance should be around 36k in cbr23 at 253w PL2. The 8 extra ecores are scoring 9k on their own.

 

[RodroX]

To bad the article forget to mention that once the damage is done to your chip, theres no turn back and that intel should accept an RMA request for it.

 

[Elusive Ruse]

The drip-feed of these mitigations doesn't inspire a ton of confidence. How do we know they aren't just pacing them out until Arrow Lake launches, at which point they'll release the final mitigation that actually has a significant performance impact.


Yeah, relative to 0x125 - and that includes running at stock power limits and Tau, which is not what a lot of gaming boards defaulted to, or how most reviewers benchmarked it.

I'd love to see how much performance the i9-14900K lost, between those original reviews and this latest update.

TH owes us a couple of new benchmarks, the post damage mitigation versions of Raptor lake and Windows 24H2 update that lifted Zen4 and 5 performance.

 

[great Unknown]

"All of the known Raptor Lake issues are now mitigated."

In my programming days, I tended to comment my code with
statements like, "This prevents such-and-such problems. FLW."

Famous Last Words.

 

[EzzyB]

The drip-feed of these mitigations doesn't inspire a ton of confidence. How do we know they aren't just pacing them out until Arrow Lake launches, at which point they'll release the final mitigation that actually has a significant performance impact.


Yeah, relative to 0x125 - and that includes running at stock power limits and Tau, which is not what a lot of gaming boards defaulted to, or how most reviewers benchmarked it.

I'd love to see how much performance the i9-14900K lost, between those original reviews and this latest update.

Probably none if you consider that stock power limits are the manufacturers recommended power limits. The fact that board makers and reviewers test things outside those limits is absolutely not Intel's fault.

It's interesting that Puget Systems, a manufacturer of high-end workstations, reported that at the same time they actually had more failures in AMD processors (though no higher than normal.) The reason? They never exceeded Intel's recommended power limits.

This whole thing smacks me like the techno-amateur with his 15% "stable overclock" (the long standing Greatest Lie in Computing) ranting about a BSOD and how Windows sucks! I mean it passed Prime 95, right?

Think about this, I'm still using a system based on a 9700K. I bought the components years ago, assembled them and, right out of the box, the processor that is supposed to run 6 of it's 8 cores at 3.6ghz base was running every core at a base of 4.8ghz with no intervention whatsoever from the user. This is because, at some point, Tom's Hardware or some other site is going to review that board and they need it to seem as fast or faster than other boards. Millions are spent because this board does 197FPS instead of that board at 194FPS and our brains can't even tell the difference!

The fact that all these mitigations are forcing the CPU to not accept out of parameter power should tell you something, right?

 

[Amdlova]

The gigabyte Z690 with the Lastest update from them.
Start the cpu with 4096w by defaults
Thank deer god the Intel has IA to limit the CPU at 84W when the PL2 is 92W

@EzzyB Intel KNOW this issue from 11 gen to today some folks push 230w - 250w on a locked i5. Intel gives that to compete with AMD and now Backfires...

On these Little T chips 50W cpu gives 90% of total chip powa! The others 10% you need increase more 50W.

14600T 1.8ghz 35w / 92W maximum clock 4.2ghz
14600K 3.5ghz 125w / 171w maximum clock 5.3ghz (STOCK) ( Intel Double the power needs for 1.1ghz Gives 10% - 15% more performance)

 

[Eximo]

 

[raysinbisket]

The fact that board makers and reviewers test things outside those limits is absolutely not Intel's fault.

Yikes. It absolutely is Intel's fault for having limits that they don't actively enforce by default. Intel has knowingly been letting this slide for years and when reviewers queried this with Intel they even said outright that exceeding the recommended limits is still considered within spec.

Now Intel will, finally, be actually enforcing the advertised limits. Took them partially bricking 1/2 generations to finally actually address this.

 

[m3city]

To bad the article forget to mention that once the damage is done to your chip, theres no turn back and that intel should accept an RMA request for it.

Cause it has been said multiple times already?

 

[logainofhades]

Probably none if you consider that stock power limits are the manufacturers recommended power limits. The fact that board makers and reviewers test things outside those limits is absolutely not Intel's fault.

Even when power limits are enforced, the CPU was asking for too much voltage. This is entirely Intel's fault.

 

[thestryker]

I'd love to see how much performance the i9-14900K lost, between those original reviews and this latest update.

Probably margin of error unless you're trying to claim reviewers who chose to run unlimited power as their only performance metric now comparing to power limited is valid.

The drip-feed of these mitigations doesn't inspire a ton of confidence. How do we know they aren't just pacing them out until Arrow Lake launches, at which point they'll release the final mitigation that actually has a significant performance impact.

This is bordering on conspiracy theory hand waving. If the mitigations were actually having any notable performance impact you might be onto something.

 

[Roland Of Gilead]

The drip-feed of these mitigations doesn't inspire a ton of confidence. How do we know they aren't just pacing them out until Arrow Lake launches, at which point they'll release the final mitigation that actually has a significant performance impact.

For sure. I've asked that very thing here a couple of times as the last few months passed. Pointing to four 'scenarios' under which the issue may occur, leaves me thinking there may be one or more 'mitigations' to come

 

[thestryker]

It isn't particularly surprising that more had to be done to stabilize the situation. Especially since they'd said there would be more after the last one. Capping the voltage requests certainly seemed like the most pressing issue given that not all affected parts even have TVB (or I'd have said this was).

At the end of the day while the motherboard manufacturers knew what they were doing configuration wise Intel did nothing to stop them. From listening to Buildzoid and reading through specifications it seems like Intel has basically designed them for worst case scenario with overly broad limits. They've needed to tighten this up for years, but it has only been exposed with pushing the boundaries of process and performance.

It's darkly humorous that heavy workloads don't really expose this issue due to lower voltage requests. Wendell from L1T talked about it on this week's PC World Full Nerd podcast explaining in terms of datacenter usage.

 

[mac_angel]

Still not going to update.
Through many days of trying to 'overclock', I finally settled on locking the voltage at 1.35V and I'm able to keep stock speeds under load without thermal throttling. That ends up being WAY faster than all these other overclocks you find online.

 

[YSCCC]

Ah ha, yet another final fix, the 0x129 was initially said to be final… now let’s see if yet another final will come..

 

[rluker5]

Still not going to update.
Through many days of trying to 'overclock', I finally settled on locking the voltage at 1.35V and I'm able to keep stock speeds under load without thermal throttling. That ends up being WAY faster than all these other overclocks you find online.

I'm sure it is also way faster than reviews where they ran the 14900k at stock settings where it was throttling.
Since you have had the vmin shift problem solved before they came out with this last microcode update, maybe reviewers should rebenchmark using your performance enhancing, degradation avoiding settings.

JK, you can't have reviewers in some never ending benchmark loop whenever anyone wants something different. People want interesting stories, not rehash after rehash.

I let my clocks and volts flop all over the place with my 13900kf, but I also keep my max voltage low.
I'm also not going to update. I might have to reenter everything in my bios saves for no gain.

With how these chips were reviewed you would think it was some cutting edge breakthrough that you could both reduce thermal/power throttling and reduce voltage induced degradation by reducing voltage.
Maybe I should keep that secret technique quiet. Who knows what could happen if word got out.

 

[toocoolblue]

1. Blame Users - Deny any problem exists.
2. Blame Motherboards - Deny any problem exists.
3. Severely reduce voltage - Deny any loss of performance exists.

It is bizarre that retail outlets such as Bestbuy and Dell are still selling systems with these chips at full price.
Considering the next gen. CPU chips - which are supposedly immune to these problems, hits shelves in two weeks.

 

[rluker5]

It is bizarre that retail outlets such as Bestbuy and Dell are still selling systems with these chips at full price.
Considering the next gen. CPU chips - which are supposedly immune to these problems, hits shelves in two weeks.

Full price still?
There must be enough people still buying them then. If nobody bought then prices would drop.

 

[TeamRed2024]

Does this issue affect mobile processors? Asking because my new laptop has the 14900HX.