News Intel's patch for CPU instability and crashing issues rolls out from MSI and ASRock — Asus ROG motherboard users can also access a beta update

[Admin]

Several Intel partners have announced or released BIOS patches to fix the erroneous microcode that leads to the instability of Raptor Lake chips.

Intel's patch for CPU instability and crashing issues rolls out from MSI and ASRock — Asus ROG motherboard users can also access a beta update : Read more

 

[Gururu]

How are sites even going to test this? What are the best controls to ensure a solid confirmation of the fix?

 

[ThisIsMe]

Always have to toss in that "at least one user claimed to have a bad experience with..." comment. Even when it's clearly just hearsay.

 

[ThatMouse]

I imagine 99% of their customers will never know about this update. BIOS updates are not something an average user should do.

 

[rene641]

What is the number of new microcode? I have a Gigabye z690 UD DDR4 and latest bios update from Gigabyte was from July 17

1. Checksum : 7298
2. Introduce the "Intel Default Settings" and enabled as default, user needs to disable it first to use GIGABYTE PerfDrive profiles.
3. Update microcode 0x125 to ensure eTVB operation normally.

 

[TJ Hooker]

I imagine 99% of their customers will never know about this update. BIOS updates are not something an average user should do.

New microcode can be applied via operating system updates, I imagine that's how most people will get this fix.

 

[JarredWaltonGPU]

How are sites even going to test this? What are the best controls to ensure a solid confirmation of the fix?

Confirmation that the 'fix' works and halts degradation will be basically impossible in the short-term. All you could do is check voltages in "unlimited" mode with an old BIOS revision and then compare it to the latest BIOS to see how things have changed.

Checking the performance impact will require using an earlier BIOS with the latest Windows patches and application updates, testing that, and then installing the latest BIOS with the fix and retesting. Any older numbers from months ago are already potentially invalid so you really need to do retesting with old and new BIOS revisions to get hard numbers that are trustworthy.

 

[jeremyj_83]

Confirmation that the 'fix' works and halts degradation will be basically impossible in the short-term. All you could do is check voltages in "unlimited" mode with an old BIOS revision and then compare it to the latest BIOS to see how things have changed.

Checking the performance impact will require using an earlier BIOS with the latest Windows patches and application updates, testing that, and then installing the latest BIOS with the fix and retesting. Any older numbers from months ago are already potentially invalid so you really need to do retesting with old and new BIOS revisions to get hard numbers that are trustworthy.

All while hoping that testing with the old BIOS doesn't wind up frying the chip.

 

[helper800]

Always have to toss in that "at least one user claimed to have a bad experience with..." comment. Even when it's clearly just hearsay.

Well considering Tomshardware is not in a court proceeding, any and all 3rd party accounts of an RMA experience is fair game for journalists, that's kinda what they do.

 

[Mattzun]

Testing voltages with an oscilloscope in scenarios that previously caused weird voltage spikes is possible.
I'm sure someone like Buildzoid will do that - he had a decent video where he showed the weird spikes that were too short to see in HWInfo.

If that works, it would show that the patch isn't a complete failure, but it wouldn't show that it actually fixes the problem(s).

The best check would be for someone who claims to be able to cause 100 percent failures within 6 months of use (i.e. Alderon games) to test it on several systems with new 14900's and release results in January.
Of course, that assumes that the patch will be available on W680 motherboards in the near future.

In any case, there is no way of telling how close to failing all of the existing, damaged CPUs are.

 

[Amdlova]

My asrock is making weird noises from vrm after bios patch, removed c6 and c7 to mitigate the noises...
Asrock h670 PG riptide

 

[Mattzun]

What is the number of new microcode? I have a Gigabye z690 UD DDR4 and latest bios update from Gigabyte was from July 17

The fix seems to be 0x129

microcode 129​

Microcode 129 is a BIOS update released by Intel and motherboard manufacturers, including ASUS and MSI, to address instability issues affecting 13th and 14th Gen Core processors, specifically the “Raptor Lake” series.
Key Features:

• Updates the microcode to version 129, which is designed to prevent degradation and resolve stability concerns
• Specifically targets the elevated operating voltage issue causing instability
• Does not affect Turbo Boost or overclocking capabilities
• Available as a beta BIOS update for select motherboards, with plans for wider distribution through OS updates (Windows and Linux)

 

[mac_angel]

how long before we can expect benchmarks?

 

[Alvar "Miles" Udell]

I doubt "enthusiasts" will notice anything since they will manually be using as much voltage as it takes to get the maximum speeds possible. People who use default or motherboard set voltages for automatic turbo, though, as well as reputable review sites, should be the ones to notice.

 

[stuff and nonesense]

How are sites even going to test this? What are the best controls to ensure a solid confirmation of the fix?

Your guess is as good as mine, once applied it will be trying to prove a negative.

The only thing I can think of, assuming 100% application of the patch, is that there will be a rapid decline in failures to near “background” levels.

 

[rluker5]

Testing voltages with an oscilloscope in scenarios that previously caused weird voltage spikes is possible.
I'm sure someone like Buildzoid will do that - he had a decent video where he showed the weird spikes that were too short to see in HWInfo.

If that works, it would show that the patch isn't a complete failure, but it wouldn't show that it actually fixes the problem(s).

The best check would be for someone who claims to be able to cause 100 percent failures within 6 months of use (i.e. Alderon games) to test it on several systems with new 14900's and release results in January.
Of course, that assumes that the patch will be available on W680 motherboards in the near future.

In any case, there is no way of telling how close to failing all of the existing, damaged CPUs are.

Not just weird voltage spikes, but also excessively high volts in general. If a chip is boosting to 1.6v on a regular basis that chip isn't going to last.

 

[Mattzun]

Not just weird voltage spikes, but also excessively high volts in general. If a chip is boosting to 1.6v on a regular basis that chip isn't going to last.

The more general high voltage issue should be easy to test - just need a bunch of people to run HWInfo in the background while using their computer.

I think Intel said the microcode should never request more than 1.55 volts
If 1.56+ volts show up as the max CPU voltage, the problem might be mitigated, but it isn't completely fixed.

 

[NightLight]

I imagine 99% of their customers will never know about this update. BIOS updates are not something an average user should do.

Neither should effing arround with cpu voltages...

 

[Mattzun]

Jay posted a video with initial test results - 2 percent performance drop in games, lower voltage spikes.
I'm sure we will see more testing soon

The final version of the microcode is supposedly going to be released through Windows update - it won't require a BIOS change.

 

[Gururu]

The more general high voltage issue should be easy to test - just need a bunch of people to run HWInfo in the background while using their computer.

I think Intel said the microcode should never request more than 1.55 volts
If 1.56+ volts show up as the max CPU voltage, the problem might be mitigated, but it isn't completely fixed.

I am curious about this. The last update (125) maxes out with ~1.45V spikes from a baseline about half of that. Those are HWINFO Vcore readings. I would like to see something different from 129.

 

[wingfinger]

It doesn't look like the fixed MC is available for linux - even as experimental.

They are only giving it to MB vendors that are in control of their customized BIOSes.

The rollout seems limited so far.

 

[Mattzun]

I am curious about this. The last update (125) maxes out with ~1.45V spikes from a baseline about half of that. Those are HWINFO Vcore readings. I would like to see something different from 129.

Jay's video showed a comparison of the spikes in 125 and 129 in a 10 min cinebench.
It looked like he had repeated spikes of up to 1.58 volts 125 and up to 1.5v in 129.

That is a significant reduction, but who knows how it will affect things if you currently max at 1.45v or if spikes to 1.45v are even an issue.

 

[Gururu]

Jay's video showed a comparison of the spikes in 125 and 129 in a 10 min cinebench.
It looked like he had repeated spikes of up to 1.58 volts 125 and up to 1.5v in 129.

That is a significant reduction, but who knows how it will affect things if you currently max at 1.45v or if spikes to 1.45v are even an issue.

Just installed the new 0x129. My Vcore spikes did not change but when I look at the Core VIDs like he did I see that the maximums have dropped. Before I would see a maximum closer to 1.47 V, now it seems fixed at around 1.447 V. I had believed Vcore was better than VID, so I was more focused on that despite Core VID being always higher. However, that is the reading that seems to have changed for me.

 

[YSCCC]

Jayz tested on his MSI platform , seems performance is more or less the same, but in between CB runs the peak reduced from ~1.58v to ~1.48, still high for my own CPU but looks ok

 

[Mindstab Thrull]

What is the number of new microcode? I have a Gigabye z690 UD DDR4 and latest bios update from Gigabyte was from July 17

It would have come out in the last couple days. Check your vendor's support website regularly for the next little while - everyone should be releasing it in the next couple weeks, and some, as mentioned, have already released for certain products.