Question Asus mobos burning x3D CpuS?

[TheELF]

New information from Paul Alcorn at Tom's Hardware.

AMD Ryzen 7000 Burning Out: EXPO and SoC Voltages to Blame (AMD Responds)

Impacts all motherboard makers and all Ryzen 7000 chips.

www.tomshardware.com

TL;DR: Industry sources say that excessive SoC voltage degrade the CPU's thermal sensors, leading to a runaway condition where the CPU draws excess power through the Vcore pins, disregarding thermal limits and killing itself.

EXPO causes the excessive SoC voltage to be set.

And we have no idea if this sensor is a new part AMD started to use and is now using in every CPU they produce,
or if it is only in the x3d chips, or the low tier ones, or what is going on.
Probably even AMD doesn't know at this point and has to investigate first.

To fully fix this issue AMD would have to find out batch numbers and do a mass recall of every affected CPU...
Anything else is just a temporary band-aid because any future bios update or even any software that can alter Vcore from windows will have the potential to burn CPUs.

 

[Kocicak]

... any future bios update or even any software that can alter Vcore from windows will have the potential to burn CPUs.

Is this not valid generally?

I am not sure if it is possible to design a CPU completely imune to any voltage motherboard can send into it.

 

[TheELF]

Is this not valid generally?

I am not sure if it is possible to design a CPU completely imune to any voltage motherboard can send into it.

The difference is that a normal CPU won't get their thermal sensor blown up and thus will stay within limits.
It will decrease power draw by decreasing clocks if Vcore increases too much which will keep it within safe temps.

You can see that here done on purpose in an attempt to blow up a CPU.

 

[CP5670]

I got the MSI beta bios and get a max VDDCR_VDD at 1.15V and VDDCR_SOC at 1.354V with Expo enabled. Not sure if that's high or not.

 

[Golgatha]

I got the MSI beta bios and get a max VDDCR_VDD at 1.15V and VDDCR_SOC at 1.354V with Expo enabled. Not sure if that's high or not.

Everything I've read so far says 1.35V or less is "safe". Also, if your thermal sensor is working and you keep your temperatures cool enough to prevent it from failing, it's also "safe".

 

[aigomorla]

Seriously: What do you do with such a puffed CPU? Do you send it to the baker (I mean MB manufacturer) for replacement, or do you send it to AMD?

Well if AMD can take that puffed dough and make it into a cream filled puff, then its a uber win in my book.

 

[Kocicak]

cream filled puff

This reminds me some naughty english word.... what is it.... oh, a different first letter...

 

[TheELF]

This reminds me some naughty english word.... what is it.... oh, a different first letter...

Muff is not a naughty word, you can say it freely.

Muff (handwarmer) - Wikipedia

en.wikipedia.org

 

[AdamK47]

This is another reason for me to be glad I know what I'm doing with manually tweaking DDR5 memory settings. EXPO is not all that it's talked up to be. This puts a massive ding to the credibility of the settings it applies.

 

[CakeMonster]

So, not an Asus issue after all then.

Just installed the 1302 BIOS. I loaded 'DOCP Tweaked' and then entered the timings I usually use. After running ZenTimings I noticed VSOC was still at 1.35V. That was a bit puzzling, because I thought these new BIOS'es were supposed to dial that back to 1.30V. Anyway, I rebooted and changed VSOC to 1.30, and while I was at it, just went for VMEM at 1.35V instead of 1.40V, its probably not even related but why not dial that back a bit since I only run at 6000, although with manual timings. I actually thought about reducing the 95C CPU temp limit too, but I couldn't find the setting and searching within the BIOS returned no changeable option, so I just left it.

One interesting finding (I've saved ZenTimings screenshots from every BIOS version upgrade) was that before 0925 VSOC was 1.25V, then it went to 1.35V, and now I've dialed it back to 1.30V, although I'd guess 1.25V should work still even though I've tightened the timings since. Maybe I'll try it next time. (This might not have been a BIOS change, its possible that I used a different DOCP profile DOCP1/DOCP2/DOCP Tweaked).

 

[CP5670]

Everything I've read so far says 1.35V or less is "safe". Also, if your thermal sensor is working and you keep your temperatures cool enough to prevent it from failing, it's also "safe".

I'll just leave it then as it has been working fine. The temps max out around 70C in Prime95. Any kind of CO causes crashes on idle though.

 

[Markfw]

This is the default bios on an ASUS prime 670E-p, except setting DOCP to DOCP I and max temp to 85C and ECO mode to 170: (looks great to me) and this is under 100% load

 

[TummyCat]

Does AMD publish some whitepapers or any other official information about their products which contain voltages and their safe ranges?




Creation of second accounts is not allowed.
Creation of second accounts under vacation/ban is a banable offense.


esquared
Anandtech Forum Director

 

[tamz_msc]

Does AMD publish some whitepapers or any other official information about their products which contain voltages and their safe ranges?

Unfortunately not, and this is one of their failings as a company when compared with Intel.

Right now, you have to take the word of strangers on what the safe voltages are.

 

[Markfw]

Unfortunately not, and this is one of their failings as a company when compared with Intel.

Right now, you have to take the word of strangers on what the safe voltages are.

AMD provides all of this information to motherboard manufacturers and they make BIOS to correctly support the CPU. Overclocking violates the warranty for BOTH AMD and Intel. You just hate AMD and take every chance you get to lie about them, and make them seem like a bad company. The rest of us know this and ignore you.

 

[tamz_msc]

AMD provides all of this information to motherboard manufacturers and they make BIOS to correctly support the CPU. Overclocking violates the warranty for BOTH AMD and Intel. You just hate AMD and take every chance you get to lie about them, and make them seem like a bad company. The rest of us know this and ignore you.

AMD: enables EXPO in marketing, supplies reviewers with EXPO kits telling them to use them in reviews.

Also AMD: Their customers wanting to enable EXPO to achieve what AMD showed are now at risk of damaging their CPUs and motherboards.

AMD stan: it's not AMD's fault, they never do anything wrong, saying negative things about them isn't allowed and is tantamount to lying.

ShockedPikachuFace.jpg

 

[Markfw]

AMD: enables EXPO in marketing, supplies reviewers with EXPO kits telling them to use them in reviews.

Also AMD: Their customers wanting to enable EXPO to achieve what AMD showed are now at risk of damaging their CPUs and motherboards.

AMD stan: it's not AMD's fault, they never do anything wrong, saying negative things about them isn't allowed and is tantamount to lying.

ShockedPikachuFace.jpg

First, EXPO is not overclocking, as you said, AMD wants people to do tis. Second, well, I have 5 7950x's, all using EXPO and have had for MONTHS, no issue.

Now we see one or 2 bad BIOS, and we don't know all the settings that were used, but since the motherboard makers came out with a fix, it was a problem they admit to.

But the post you replied to was about do USERS know these voltages. THEY DON'T NEED TO UNLESS THEY ARE OVERCLOCKING.

And last, the whole issue has virtually nothing to do with AMD. The motherboard makers had a problem in their BIOS, and they admitted it. But again, you mock AMD and blame them.

 

[TheELF]

And last, the whole issue has virtually nothing to do with AMD. The motherboard makers had a problem in their BIOS, and they admitted it. But again, you mock AMD and blame them.

It might turn out to be something else in the end, but the way it looks now is that AMD used a bad thermal sensor that can't withstand enough Voltage to be considered decent. If your protection mechanism only goes up to 1.35V then this is a serious issue from AMD and everything else is just a smoke screen, if the CPU were ok and well made then setting overclocking voltages would just be ignored by the CPU, it would go up to what the CPU can stand and would stop there.
But here the protection burns out and not at some extreme high Voltage but at voltage that you could get just randomly at boot up and that causes the CPU to go higher and higher until it blows.

AMD Ryzen 7000 Burning Out: EXPO and SoC Voltages to Blame (AMD Responds)

Impacts all motherboard makers and all Ryzen 7000 chips.

www.tomshardware.com

Our sources also added further details about the nature of the chip failures — in some cases, excessive SoC voltages destroy the chips' thermal sensors and thermal protection mechanisms, completely disabling its only means of detecting and protecting itself from overheating. As a result, the chip continues to operate without knowing its temperature or tripping the thermal protections.

 

[coercitiv]

First, EXPO is not overclocking, as you said, AMD wants people to do tis.

EXPO is overclocking, don't pick this hill.

Here's the footnotes AMD provides for their marketing slogan for EXPO:

1. Overclocking and/or undervolting AMD processors and memory, including without limitation, altering clock frequencies / multipliers or memory timing / voltage, to operate outside of AMD’s published specifications will void any applicable AMD product warranty, even when enabled via AMD hardware and/or software. This may also void warranties offered by the system manufacturer or retailer. Users assume all risks and liabilities that may arise out of overclocking and/or undervolting AMD processors, including, without limitation, failure of or damage to hardware, reduced system performance and/or data loss, corruption or vulnerability. GD-106

 

[mv2devnull]

That is no different from Intel processor's Limited Warranty statement:

does NOT cover:
* any Product that has been modified or operated outside of Intel’s publicly available specifications, including where clock frequencies or voltages have been altered with any Intel tools or through other means; Intel does not give or enter into any condition, warranty, or other terms to the effect that the Product, including if used with altered clock frequencies or voltages, (1) will be fit for any particular purpose, or (2) will not cause any damage or injury.

and explanation:

Usage of overclocking and Intel® XMP and warranty Info
Altering clock frequency or voltage may void any product warranties and reduce stability, security, performance, and life of the processor and other components. Check with system and component manufacturers for details.

 

[Markfw]

EXPO is overclocking, don't pick this hill.

Here's the footnotes AMD provides for their marketing slogan for EXPO:

Getting awful technical there. First, wiki definition "In computing, overclocking is the practice of increasing the clock rate of a computer to exceed that certified by the manufacturer."

So AMD designed EXPO memory. The motherboard manufacturer, in support of this design, includes a setting that uses the MEMORY MANUFACTURER'S design frequency specification. So if you use that, you are not exceeding the design spec. Also, right at the bottom of the AMD spec page is this:

" Supported Technologies AMD EXPO™ Technology"

But then AMD says their chip runs at DDR-5200 max listed. So which is it ? You can can make the argument either way IMO. BUT when you consider "excessive SoC voltages destroy the chips' thermal sensors" then you definitely are going outside then spec.

But again, my response was to a user saying that AMD had to provide voltage information to the user, and then a reply to that, which is NOT REQUIRED if you are not overclocking was my point.

 

[tamz_msc]

But again, my response was to a user saying that AMD had to provide voltage information to the user, and then a reply to that, which is NOT REQUIRED if you are not overclocking was my point.

Intel provides datasheets, regardless of whether the end-user is overclocking or not.

With AMD: What's the safe voltage for the memory controller? Answer: Nobody knows; here's some extreme overclocker sharing his opinion on YouTube.

Same question with Intel: Sure, look up page no. xyz in the datasheet available on the website.

See? The difference in approach of the two companies matters to the end-user who is interested in performance tuning.

Also, this fiasco validates the stance of Anandtech resolutely refusing to test at anything beyond max. JEDEC supported memory on either platforms.

Also, if all reviewers now start testing @ JEDEC speeds, it will make AMD look bad because default memory support on Intel is 1. 400 MHz higher, 2. Independent testing has confirmed that AMD loses a lot of performance with default memory, more compared with Intel.

 

[DrMrLordX]

New information from Paul Alcorn at Tom's Hardware.

AMD Ryzen 7000 Burning Out: EXPO and SoC Voltages to Blame (AMD Responds)

Impacts all motherboard makers and all Ryzen 7000 chips.

www.tomshardware.com

TL;DR: Industry sources say that excessive SoC voltage degrade the CPU's thermal sensors, leading to a runaway condition where the CPU draws excess power through the Vcore pins, disregarding thermal limits and killing itself.

EXPO causes the excessive SoC voltage to be set.

That's interesting. And not something that would be easily diagnosed just poking at dead hardware.

This is another reason for me to be glad I know what I'm doing with manually tweaking DDR5 memory settings. EXPO is not all that it's talked up to be. This puts a massive ding to the credibility of the settings it applies.

Yeah ditto, I never did like XMP and I'm kinda glad my kit isn't EXPO either. It'll take me some time to get this RAM running right, but the vSoC values people are reporting with EXPO are kinda nuts. Why is vSoC getting so high all of a sudden? It wasn't even supposed to go past ~1.2v on AM4 for most CPUs.

 

[coercitiv]

Sure, here's one for the Rocket Lake platform:

From page 125 of this document:

https://cdrdv2.intel.com/v1/dl/getContent/634648

I was going to drop this considering it's an AMD thread, but you keep mentioning intel's solid documentation, so I guess my reply is due.

I asked for "safe voltage" as in the max value, not the typical voltage. Here's the typical VDDQ supplied by Intel for Alder Lake, the same 1.2V you found for RKL. And here's what my motherboard supplies in Auto mode for VDDQ: 1.3V.

Spoiler: UEFI Screen

So what do you think it's the safe memory controller value on the Intel side? The 1.2V Intel mentions... or the 1.3V MSI uses on Auto... or the 1.4+V that people sometimes use to reach the higher range of overclocked DDR4 results such as DDR4 4000+ ?



Both Intel and AMD chose to take a hands-off approach to memory overclocking through XMP/EXPO. There are no guidelines from them as to what voltages to use. Users either rely on auto motherboard magic or scour the overclocking resources for word of mouth on what parameters scale their OC and/or what are safe values (either short or long term).

I appreciate Intel's datasheets, I relied on them many times already, but they are not the solution needed here.

 

[tamz_msc]

I asked for "safe voltage" as in the max value, not the typical voltage.

It says right there that VDDQ_Max is Typ. + 5%, so that'll be 1.26 V. Note that the terminology changes from Rocket Lake to Alder Lake, so VDDQ in the former refers to MC supply voltage, while in the latter it refers to DRAM voltage. So it is obvious that it is meaningless to give a max value for the latter, as it will depend on the memory speed.