AMD Releases First Statement on Ryzen 7000X3D Series Burn-out Issues

[btarunr]

AMD late Tuesday released its first media statement on the controversy surrounding Ryzen 7000X3D series processors burning-out as a result of voltage-assisted overclocking. We've covered this in detail, in our older article. The AMD statement reads that the company is aware of the issue, is investigating it, and in the meantime, is getting motherboard- and ODM vendors to ensure that their device firmware/BIOS operate Ryzen 7000X3D processors within the correct voltage tolerances. The statement also called for affected users to reach out to AMD Support.

"We are aware of a limited number of reports online claiming that excess voltage while overclocking may have damaged the motherboard socket and pin pads. We are actively investigating the situation and are working with our ODM partners to ensure voltages applied to Ryzen 7000X3D CPUs via motherboard BIOS settings are within product specifications. Anyone whose CPU may have been impacted by this issue should contact AMD customer support."

The AMD statement follows several of its motherboard partners pushing out UEFI firmware (BIOS) updates, and some even removing older versions of BIOS from their Support websites. The new firmware enforces strict limits on CPU core voltages, and prevents voltage-assisted overclocking. Some companies, such as MSI, even introduced new automated overclocking modes that enhance PBO-based performance tuning (improved boost frequency residency), without crossing the voltage limits set by AMD.

Many Thanks to DeathtoGnomes for the tip.

View at TechPowerUp Main Site | Source

 

[Jism]

It's so stupid. Someone must have thought, oh there might be a percentage of CPU's that maybe could not run at the desired speeds of over 6000Mhz. So lets raise the SOC voltage to a scorching 1.4V to guarantee that all chips should work on the desired settings.

And actually not even thinking of what could happen, because yeah 1.45v should be safe right? Soc voltage of beyond 1.3V was never considered safe. 1.2V is still in my opinion far beyond what is normal.

 

[Daven]

AMD and motherboard companies are moving fast and on this even though only one (?) case has been confirmed. AMD CPUs with X3D represent the highest tech currently available for client computing. I’m sure AMD wants to get this absolutely right.

 

[Naito]

I'd be inclined to point the finger at the board manufacturers. Always trying to out do each other with clocks for marketing. Wouldn't be surprised if they were a bit liberal with voltages

 

[TumbleGeorge]

All is guilty for trying to steal more money from costumers!

 

[Jism]

AMD and motherboard companies are moving fast and on this even though only one (?) case has been confirmed. AMD CPUs with X3D represent the highest tech currently available for client computing. I’m sure AMD wants to get this absolutely right.

And AMD deliveres quality. It started from the AM4 platform with the first ryzen already. There where no motherboards to be seen that would chrash, throttle or whatever due to the high VRM requirement. Even 40$ of a board is capable of running a 5950X even in OC'ed state. AMD gave motherboard makers too much of a freedom with it's AM3 and caused quite a bad name.

This is a misconfiguration coming from EXPO - a setting that dials the SOC voltage beyond 1.4V which is completely unneeded, and yet fries the CPU + socket with it. The bios releases as of now cap that SOC voltage limit back to 1.3V max.

 

[Outback Bronze]

I've been reading/hearing this problem is not just limited to the X3D's either. I wonder why they're commenting about the X3D's only?

 

[phanbuey]

I've been reading/hearing this problem is not just limited to the X3D's either. I wonder why they're commenting about the X3D's only?

Probably just information lag - statements like this take time coming out of larger companies.

As they have to acknowledge the issue without panicking the shareholders.

 

[Dragam1337]

AMD and motherboard companies are moving fast and on this even though only one (?) case has been confirmed. AMD CPUs with X3D represent the highest tech currently available for client computing. I’m sure AMD wants to get this absolutely right.

No, there has been alot of examples from repair shops seeing this exact damage...

I've been reading/hearing this problem is not just limited to the X3D's either. I wonder why they're commenting about the X3D's only?

If i were to guess, it's cause that while all ryzen 7000 cpu's potentially have this issue, it's exacerbated on the 3d chips.

 

[Outback Bronze]

Probably just information lag - statements like this take time coming out of larger companies.

As they have to acknowledge the issue without panicking the shareholders.

The funny thing is I remember when the X3D's were first getting released and even TPU had a news article about how they were going to be "overclockable".

Now every board manufacturer has taken their OC bios's offline with new ones that don't support OC. WTF hey.

 

[dgianstefani]

AMD and motherboard companies are moving fast and on this even though only one (?) case has been confirmed. AMD CPUs with X3D represent the highest tech currently available for client computing. I’m sure AMD wants to get this absolutely right.

If they wanted to get it absolutely right they could try communicating with their board partners ahead of release rather than months after release, only after several critical failures that resulted in killing not just the CPUs but also the board they are in.

 

[bug]

I'd be inclined to point the finger at the board manufacturers. Always trying to out do each other with clocks for marketing. Wouldn't be surprised if they were a bit liberal with voltages

People also point fingers at Intel or Nvidia for maintaining stricter control over partners' designs. Yet if you don't do that, this happens...

 

[swirl09]

I've been reading/hearing this problem is not just limited to the X3D's either. [...]

Der8auer has a video up about this, seems it can affect any 7000 chip. He said some posts on reddit incl people who did not OC their chips (Im pressing X on that claim for now.)

 

[Dragam1337]

The funny thing is I remember when the X3D's were first getting released and even TPU had a news article about how they were going to be "overclockable".

Now every board manufacturer has taken their OC bios's offline with new ones that don't support OC. WTF hey.

I don't think many expected that they would actually be able to oc it, other than using pbo and expo - i certainly didn't anyways.
And i would personally much rather have a safe chip, than a few % more performance.

 

[Xuper]

I'm surprised they responded Fast based on one sample!

 

[phanbuey]

I don't think many expected that they would actually be able to oc it, other than using pbo and expo - i certainly didn't anyways.
And i would personally much rather have a safe chip, than a few % more performance.

The crazy thing is motherboards have been doing this forever. My soc was running at 1.39v at 6800 64gb pre last bios update — now it’s at 1.19 with the same settings.

auto motherboard settings have always been notorious overvolters for zen— even at stock settings. But those chips can generally handle high voltages - to have enough voltage / heat to warp silicon like that with modern Ovp and thermal protection there has to be something else going on - that hotspot had to be at least 150-200C, there is way more to this that we are about to find out.

 

[Hyderz]

Well the the overclock tests show you shouldn’t really push the cpu much further, also the x3d is already a blazing fast cpu beating out the 13900k/ks in gaming.

 

[phanbuey]

Well the the overclock tests show you shouldn’t really push the cpu much further, also the x3d is already a blazing fast cpu beating out the 13900k/ks in gaming.

Not at stock ram settings tho -- all the tests where it crushes intel is with 6000 CL30 ram even from amd. If you put it to 5200 stock ram vs 13900K at 6000 it loses the lead and they're basically tied:

As soon as you start yoking subtimings motherboard is liable to start pumping voltages to the soc, so you would basically have to run 5200CL40 to be completely stock.

Put a ring OC and some extra MHZ to the 13900K and it actually beats.

 

[Daven]

No, there has been alot of examples from repair shops seeing this exact damage...

So far only one confirmed example. Saying there are a lot of examples is hearsay at best. Remember this is the internet. We don’t know each other and Daven and Dragam1337 are not our real names.

If you have a link to articles from reputable sources citing evidence from repair shops then please provide as many are very interested in whats going on here including me. Otherwise none of us can just accept what you say.

 

[Ellothere]

Not at stock ram settings tho -- all the tests where it crushes intel is with 6000 CL30 ram even from amd. If you put it to 5200 stock ram vs 13900K at 6000 it loses the lead and they're basically tied:

View attachment 293219

As soon as you start yoking subtimings motherboard is liable to start pumping voltages to the soc, so you would basically have to run 5200CL40 to be completely stock.

Put a ring OC and some extra MHZ to the 13900K and it actually beats.

So your comparing 13900k with fastest ram vs 7950x3d (which 7800x3d is faster) with slowest….. looks like across the board like for like amd is faster.

 

[bug]

I'm surprised they responded Fast based on one sample!

One sample that we know of.

 

[TumbleGeorge]

If rumours is not rumours we have first decreasing of prices of 7000X3D series.

 

[evernessince]

This is a misconfiguration coming from EXPO - a setting that dials the SOC voltage beyond 1.4V which is completely unneeded, and yet fries the CPU + socket with it. The bios releases as of now cap that SOC voltage limit back to 1.3V max.

Incorrect, an EXPO memory profile, just like an XMP memory profile, does not contain an SoC value. The SoC value is set by the motherboard.

If they wanted to get it absolutely right they could try communicating with their board partners ahead of release rather than months after release, only after several critical failures that resulted in killing not just the CPUs but also the board they are in.

The 7000 series has been out since October 2022, that's we are only hearing of this now hints at something else other than bad launch communication regarding bugs.

Not at stock ram settings tho -- all the tests where it crushes intel is with 6000 CL30 ram even from amd. If you put it to 5200 stock ram vs 13900K at 6000 it loses the lead and they're basically tied:

View attachment 293219

Neither processors are going to be running stock SoC or RAM voltages at their top memory speeds so that's really a moot point.

As soon as you start yoking subtimings motherboard is liable to start pumping voltages to the soc, so you would basically have to run 5200CL40 to be completely stock.

This is just false. The motherboard sets a fixed SoC voltage when you enable EXPO. Further tuning sub-timings will not impact that fixed voltage.

Put a ring OC and some extra MHZ to the 13900K and it actually beats.

Irrelevant to the topic. We get it, if you handicap the Zen processor, OC only the 13900K, and use 320w more watts eventually the 13900K will beat AMD's 2nd fastest gaming processor. Cool story and it still gets beaten by the 7800X regardless.

 

[phanbuey]

So your comparing 13900k with fastest ram vs 7950x3d (which 7800x3d is faster) with slowest….. looks like across the board like for like amd is faster.

Yeah if you can't use EXPO because the chips are exploding - and you're stuck at STOCK 5200; then your top speed is the stock non-expo speed.

 

[evernessince]

Yeah if you can't use EXPO because the chips are exploding - and you're stuck at STOCK 5200; then your top speed is the stock non-expo speed.

This is called an over-reaction. I've had a 7700X since launch October 2022 with EXPO enabled and zero issues. Pin pads still in perfect condition.

Your scare-mongering will age like those jumping to conclusions about the 4000 series power adapters, poorly