anandtech MSI implements voltage lock in new BIOS updates to protect Ryzen 7000X3D processors ( AMD UPDATE )

[bonami2]

Quote

In response to recent reports of AMD's Ryzen 7000X3D processors being damaged, MSI has announced a wave of firmware updates to address potential issues. The highlight of MSI's new wave of BIOS versions for its AM5 motherboards is that the company is further locking down the voltage controls for X3D chips, removing support for positive offset voltages. Going forward, it will only be possible to use negative offset voltages on X3D chips, ensuring that the sensitive V-Cache-equipped X3D chips can't be overvolted and put at risk of damage.

https://www.anandtech.com/show/18825/msi-addresses-cpu-voltages-on-am5-motherboards-for-ryzen-7000x3d-processors

 

New Update here : https://videocardz.com/newz/amd-identifies-the-root-cause-of-ryzen-7000-cpus-burnout-issue-new-agesa-released

Edited May 3, 2023 by bonami2

[ENTERPRISE]

I see both sides to this. While I understand that AMD is very likely leaning on manufacturers to limit the potential of damage to the CPU and by extension RMA's, it also sucks for those who opted for one of these CPU's will now lose an avenue of chasing extra performance. 

 

However, with PBO being what it is these days, for those wanting the performance of the X3D chips, the loss of overclocking abilities may be a worthwhile trade off for most who are happy to just let PBO run its course.

[J7SC_Orion]

...positive offset voltages should never have been offered in general bios in the first place, IMO, but especially not with the heat-sensitive 3D models. Positive offset bios should only be available for select XOC/ sub-zero benchers. Negative offset can work quite well. I'm glad though that I decided to wait with upgrading from my 5950X...the RTX 4K goodies such as DLSS3, FrameGen and NVReflex 'upgraded' the whole system in the main games and apps I use. My 5950X on Asus DarkH 570 is now two years old and still performs exactly as on day one; never used positive offset, even with benching (below, spoiler) on extensive w-cooled setup.

 

In any case, when Intel's X99 chips first came out, Asus had 'early bios' problems (VCCSA 'auto' way too high) that fried CPUs, including one of mine which fortunately was replaced by the powers that be. Part of the problem were those 'unofficial' extra pins that were enabled by Asus...

 

Spoiler

 

Edited April 25, 2023 by J7SC_Orion

[J7SC_Orion]

...sorry for the consecutive posts; just saw this:

 

 

 

...& not sure if this one had already been posted at EHW..

 

 

Edited April 26, 2023 by J7SC_Orion

[J7SC_Orion]

...now we're getting somewhere on the mystery-melt. Very detailed 38 min; if you run a Ryzen 7K, you should probably watch this...

 

 

[ENTERPRISE]

On 30/04/2023 at 20:46, J7SC_Orion said:

...now we're getting somewhere on the mystery-melt. Very detailed 38 min; if you run a Ryzen 7K, you should probably watch this...

 

 

 

That was entertaining!

[Snakecharmed]

Asus capped the SOC voltage to 1.30 V in their latest BIOSes for all Ryzen 7000 CPUs, not just the X3Ds, but it was still a bit over in HWiNFO on my motherboard, reading as high as 1.328. I manually lowered it to 1.25 and now it settles in at 1.272. It's still way over what other OEMs are doing with their AMD EXPO-enabled vSOC settings though.

 

Before I updated the BIOS, I had the vSOC on Auto. The BIOS identified it as 1.32 V and it read as high as 1.38 in HWiNFO. That's pretty ridiculous and reckless on Asus's part considering other manufacturers were running around 1.20 V with EXPO enabled.

Edited May 3, 2023 by Snakecharmed

[bonami2]

Update: https://videocardz.com/newz/amd-identifies-the-root-cause-of-ryzen-7000-cpus-burnout-issue-new-agesa-released

[neurotix]

Updated my board today and yes, Asus has implemented a voltage limit of 1.3v SOC voltage. I already had mine set to 1.25v manually for the entire time I have had my setup.

 

System also posts way way faster, memory training happens a lot faster (like 5 seconds instead of 30)

 

Unfortunately, I've run into the issue with my 2nd CCD not boosting past 5.2GHz in Performance power plan or even 1usmus Ryzen Optimized Balanced power plan. A bios update fixed this for me, and since I bench for HWBOT and finally found the fault with my 4090 (replaced the PSU), I went to try benching today and found out it won't boost past 5150MHz on my 2nd ccd. Last time I tried in 3dmark11 it was boosting to 5625MHz. Not sure what I'm doing wrong, nothing has changed in the bios settings yet chip won't boost on power plans other than Balanced (which only boosts the first, slower CCD).  They really need to fix this permanently because it's not the first time and a bios update fixed it before, now before I updated today it wasn't boosting high anymore. Updating the bios today did not fix the issue.

Edited May 3, 2023 by neurotix

[J7SC_Orion]

As mentioned before, this seems to be almost a carbon-copy of the early X99 VCCSA over-voltage in auto mode - mostly in early Asus bios - back in the day...almost 50% over stated Intel guidelines in auto mode, and that killed a few CPUs.