• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Ryzen 7 7800X3D CPU Burns Up

Soldato
OP
Joined
12 Feb 2014
Posts
2,864
Location
Somewhere Only We Know
What voltage is your VDDR_SOC at on the latest BIOS out of interest?
I've manually set it to 1.15v, on auto it still wants to do 1.35v, I'm going to bring the mem controller voltage down tonight which is on 1.35v too, and test, see if I can get that down to 1.2v / 1.25v, I don't think either of these voltages need to be anywhere near as high as they are.
 
Soldato
Joined
23 Jul 2007
Posts
2,823
Location
Worcester
I've manually set it to 1.15v, on auto it still wants to do 1.35v, I'm going to bring the mem controller voltage down tonight which is on 1.35v too, and test, see if I can get that down to 1.2v / 1.25v, I don't think either of these voltages need to be anywhere near as high as they are.
Cool, yeah mine are both at 1.2v stable as it stands, I guess I could try reducing the SOC down a bit more too. I did get a crash last night with them both at 1.15v, but I'm not sure if that was due to one or the other, and missed the actual crash as I was downstairs making a drink lol.
 
Soldato
OP
Joined
12 Feb 2014
Posts
2,864
Location
Somewhere Only We Know
Cool, yeah mine are both at 1.2v stable as it stands, I guess I could try reducing the SOC down a bit more too. I did get a crash last night with them both at 1.15v, but I'm not sure if that was due to one or the other, and missed the actual crash as I was downstairs making a drink lol.
Thanks for the confirmation, at least I know what to shoot for, I've read today another guy manually set both to 1.25v and left them there.

It's a bit bad really as someone who doesn't know much about a bios but knows how to put a machine together is just going to leave this all on auto and potentially end up with a dead system because of it over volting and no fault of their own.
 
Associate
Joined
6 Apr 2023
Posts
46
Location
UK
My SoC voltage is at 1.15v - no issues with running @ 6000mhz cl30 and very tight subtimings. Have run Prime95 large FFT's and its rock solid.
I've not lowered my VDDIO voltage, default for AM5 CPU's is 1.2v (at least on this motherboard) and I'm far less concerned with EXPO raising it to 1.35v.

It was the significant power increase with SoC that concerned me and there is no appreciable power increase that can be seen with package power with the VDDIO voltage bumped up to 1.35v
 
Last edited:
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
Mobo vendors push SoC voltage and sa voltage on amd and Intel respectively for compatibility reasons. It makes sense, by pushing the voltages they make sure that your ram kit will work with expo / xmp even if you end up with a very bad bin.

Put SoC voltages still don't explain the socket burning, at all. The CPU can die due to excess voltage, but frying itself is a different thing all together. I have to assume that the cpu protections are dying first due to the SoC voltage and then the cpu draws unlimited amounts of power (over 9 thousand) which results in a burned socket.
 
Associate
Joined
6 Apr 2023
Posts
46
Location
UK
Mobo vendors push SoC voltage and sa voltage on amd and Intel respectively for compatibility reasons. It makes sense, by pushing the voltages they make sure that your ram kit will work with expo / xmp even if you end up with a very bad bin.

Put SoC voltages still don't explain the socket burning, at all. The CPU can die due to excess voltage, but frying itself is a different thing all together. I have to assume that the cpu protections are dying first due to the SoC voltage and then the cpu draws unlimited amounts of power (over 9 thousand) which results in a burned socket.

That is the theory - SoC overvoltage is killing part of the IO chip that monitors and controls the power ramping of the CPU based on load. Zen 4 chips ramp up power, under the boost strategy, until they hit limiters (thermal, voltage, power etc). If nothing is monitoring the limiters (as it's burnt out) then....
The chip ramps up power uncontrollably (i.e. requests higher and higher VDDCR voltages from the board VRM's) - this causes the CCD chiplets to burn out, short, and get so hot they burn the substrate of the CPU, which in turn cooks the pins and socket in that area where the CCD is.

To sum up........SoC overvoltage kills internal CPU monitoring -> MB overvolts CPU core on request of the CPU -> CPU incinerates itself -> CPU burns the MB socket
 
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
That is the theory - SoC overvoltage is killing part of the IO chip that monitors and controls the power ramping of the CPU based on load. Zen 4 chips ramp up power, under the boost strategy, until they hit limiters (thermal, voltage, power etc). If nothing is monitoring the limiters (as it's burnt out) then....
The chip ramps up power uncontrollably (i.e. requests higher and higher VDDCR voltages from the board VRM's) - this causes the CCD chiplets to burn out, short, and get so hot they burn the substrate of the CPU, which in turn cooks the pins and socket in that area where the CCD is.

To sum up........SoC overvoltage kills internal CPU monitoring -> MB overvolts CPU core on request of the CPU -> CPU incinerates itself -> CPU burns the MB socket
Oh, okay, yeah that would make sense then
Wow!
 
Soldato
Joined
1 Nov 2002
Posts
10,243
Location
Sussex
Put SoC voltages still don't explain the socket burning, at all. The CPU can die due to excess voltage, but frying itself is a different thing all together. I have to assume that the cpu protections are dying first due to the SoC voltage and then the cpu draws unlimited amounts of power (over 9 thousand) which results in a burned socket.

I did read something earlier that says that’s pretty much exactly what’s happening (from the Tom’s link).

‘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.’
 
Associate
Joined
6 Apr 2023
Posts
46
Location
UK
From what I gather, so far all the current motherboard BIOS updates seem to have done is potentially save themselves (but the CPU is toast) - it looks like they have implemented VRM & thermal limits on the motherboard - so regardless of what the CPU requests for it's VDDCR voltage, under it's death condition - the board will not supply more than a pre-determined voltage (probably 1.2v max on x3d). Until they have sorted out what they are doing with EXPO, the latest bios on my ASUS board at least is still pumping out an unsafe 1.35v for SoC when EXPO is enabled.

The above changes however mean the CPU will not go into complete meltdown, but it will still die (SoC overvoltage will still kill various IO core components including the internal monitoring - just the board will no longer supply the lethal amount of voltage for it to go on and burn the whole chip and socket with it).
 
Last edited:
Soldato
Joined
8 Oct 2009
Posts
3,224
Location
United Kingdom
On an MSI B650 Carbon and 7800X3D. Ive reduced my VDDCR SOC voltage from 1.35v to a more sensible 1.20v in the hopes that I've not damaged it/will not damage it :D
Truth be told i have no idea what I'm doing but from what I'm reading 1.15-1.20 with EXPO seems to be a stable sweet spot until we know more (or just disable EXPO all together).
 
Last edited:
Soldato
OP
Joined
12 Feb 2014
Posts
2,864
Location
Somewhere Only We Know
That is the theory - SoC overvoltage is killing part of the IO chip that monitors and controls the power ramping of the CPU based on load. Zen 4 chips ramp up power, under the boost strategy, until they hit limiters (thermal, voltage, power etc). If nothing is monitoring the limiters (as it's burnt out) then....
The chip ramps up power uncontrollably (i.e. requests higher and higher VDDCR voltages from the board VRM's) - this causes the CCD chiplets to burn out, short, and get so hot they burn the substrate of the CPU, which in turn cooks the pins and socket in that area where the CCD is.

To sum up........SoC overvoltage kills internal CPU monitoring -> MB overvolts CPU core on request of the CPU -> CPU incinerates itself -> CPU burns the MB socket
This makes perfect sense, let just hope we've not already all killed the protection section in the io die due to the soc over voltage, otherwise AMD are going to get a right pile of chips back that they're not going to be able to ignore.
 
Soldato
Joined
31 Oct 2002
Posts
9,952
Yikes, terrible news for AMD. This shouldn't happen - rushing products to market and not testing for long enough is responsible.

I've swapped my 4090 over to my 13900k system and shall use that for now. I'll wait until AMD/ASUS release fully updated BIOS's for my 7950X3D/X690E Hero and 6000Mhz C30 EXPO kit before powering them on again. Aware I could manually lower voltages, but don't have time to fiddle/stability test with undervolts with work demands currently.

7950X3D is a fantastic CPU overall though, outclasses Intel with extremely good efficiency and performance, but this will hurt AMD's mindshare/public view.
 
Last edited:
Associate
Joined
8 Sep 2020
Posts
1,460
This was just posted on the ASUS PC DIY FB group .

ASUS PC DIY Group | Facebook

This group is not an official support channel or forum. Welcome to the ASUS PCDIY Facebook Group! – This group is all about PC building & PC DIY. Whether it be a new build, an upgrade, or a mod you...
www.facebook.com
www.facebook.com

Community Notice - ASUS AM5 motherboard owners

An important update for Ryzen7000X3D processor owners

Several AMD Ryzen 7000X3D owners have reported CPU and motherboard failures. We acknowledge the incidents/issues and have been communicating with AMD to analyze the possible causes. We have also contacted affected users to provide support and collect additional information.
Ryzen 7000X3D processors do not allow for CPU ratio or CPU core voltage tuning (CPU overclocking) but do allow for performance tuning and DRAM overclocking via PBO2 and EXPO memory. To support EXPO and/or memory overclocking at DDR5-6000 and beyond, SoC voltage has to be sufficiently increased to ensure compatibility and stability. The amount of voltage required varies between CPU samples. Some processors are more sensitive to overvoltage than others, and some are capable at running higher memory frequencies without needing as much voltage.
As confirmed with AMD, any intentional manipulation of these settings can damage the processor, socket, and motherboard. To mitigate this, we have been working with AMD to define new rules for EXPO memory and SoC voltage. To help protect the CPU and motherboard, we are issuing new EFI updates to limit the maximum available SoC voltage to 1.3V.
We recommend updating your motherboard UEFI BIOS to the latest release. Please also ensure the CPU is cooled adequately. Our recommendation is to use at least a 240mm AIO liquid cooler or high-performance air cooler.
If you have been affected, please do not hesitate to contact ASUS support for your region.

Any additional updates will be noted and updated in this post.

FAQ -

1. When are the new UEFI releases going to be released. Our expectation is to have the UEFI BIOS updates posted and available through the service and support website within the next 24 hours.
 
Soldato
OP
Joined
12 Feb 2014
Posts
2,864
Location
Somewhere Only We Know
Yikes, terrible news for AMD. This shouldn't happen - rushing products to market and not testing for long enough is responsible.

I've swapped my 4090 over to my 13900k system and shall use that for now. I'll wait until AMD/ASUS release fully updated BIOS's for my 7950X3D/X690E Hero and 6000Mhz C30 EXPO kit before powering them on again. Aware I could manually lower voltages, but don't have time to fiddle/stability test with undervolts with work demands currently.

7950X3D is a fantastic CPU overall though, outclasses Intel with extremely good efficiency and performance, but this will hurt AMD's mindshare/public view.

Not long to wait, new bios for ASUS motherboards with soc voltage protection, should be on the site in the next day or 2 https://docs.google.com/spreadsheets/d/19gqxeeAdUmkwgE0EvH4MlJ7meeDbEHYP-FHYW-PMyOk/edit#gid=5220165...
 
Last edited:
Associate
Joined
6 Apr 2023
Posts
46
Location
UK
This was just posted on the ASUS PC DIY FB group .

ASUS PC DIY Group | Facebook

This group is not an official support channel or forum. Welcome to the ASUS PCDIY Facebook Group! – This group is all about PC building & PC DIY. Whether it be a new build, an upgrade, or a mod you...
www.facebook.com
www.facebook.com

Community Notice - ASUS AM5 motherboard owners

An important update for Ryzen7000X3D processor owners

Several AMD Ryzen 7000X3D owners have reported CPU and motherboard failures. We acknowledge the incidents/issues and have been communicating with AMD to analyze the possible causes. We have also contacted affected users to provide support and collect additional information.
Ryzen 7000X3D processors do not allow for CPU ratio or CPU core voltage tuning (CPU overclocking) but do allow for performance tuning and DRAM overclocking via PBO2 and EXPO memory. To support EXPO and/or memory overclocking at DDR5-6000 and beyond, SoC voltage has to be sufficiently increased to ensure compatibility and stability. The amount of voltage required varies between CPU samples. Some processors are more sensitive to overvoltage than others, and some are capable at running higher memory frequencies without needing as much voltage.
As confirmed with AMD, any intentional manipulation of these settings can damage the processor, socket, and motherboard. To mitigate this, we have been working with AMD to define new rules for EXPO memory and SoC voltage. To help protect the CPU and motherboard, we are issuing new EFI updates to limit the maximum available SoC voltage to 1.3V.
We recommend updating your motherboard UEFI BIOS to the latest release. Please also ensure the CPU is cooled adequately. Our recommendation is to use at least a 240mm AIO liquid cooler or high-performance air cooler.
If you have been affected, please do not hesitate to contact ASUS support for your region.

Any additional updates will be noted and updated in this post.

FAQ -

1. When are the new UEFI releases going to be released. Our expectation is to have the UEFI BIOS updates posted and available through the service and support website within the next 24 hours.

1.3v still too high imo, but it's a lot better than the 1.4v+ I've seen some of the ASUS ROG series boards put out on SoC with EXPO.
 
Associate
Joined
8 Sep 2020
Posts
1,460
1.3v still too high imo, but it's a lot better than the 1.4v+ I've seen some of the ASUS ROG series boards put out on SoC with EXPO.
I'm on Intel Z790 Hero but its identical what they do with intel and left to auto ramps things way to high imo :eek: must get in that bios and manual tune things , luckily their bios is easy enough to navigate around :D
 
Back
Top Bottom