Question Asus mobos burning x3D CpuS?

[MarkPost]

?

 

[TheELF]

They gave voltages of their MBs for comparison, wich allow to check alleged veracity of faulty bioses, on the other end you re just here to badmouth AMD as aknowledged by your track record.

Edit : What we can read here and there is that AMD has provided bioses blocks that allow to significantly outspec the standard settings, Asus among others is well known to push the specs to grab a handfull percent performance.

Yes, the few that had CPUs that blew up...



but how is everybody else with an AMD CPU that still works different from anybody with an intel CPU?

Also the more you troll me with statements like that the more I'm going to play your game, just saying, If you want civil conversations don't call people names.
Just because you are butthurt about your precious AMD doesn't mean that everybody else is trying to badmouth them.

 

[Abwx]

but how is everybody else with an AMD CPU that still works different from anybody with an intel CPU?

Dunno, ask member Hulk when it comes to your revered brand, he will give you some hints about his previous stock set CPU...

Back to topic one should know that AMD doesnt provide an unified bios, they provide different blocks that are then stuck together by the MB manufacturer depending of the MB features, that s why there s no two same bioses from one brand to another and even within a same brand.

Some brands will try to maximise the perfs by often out the box overclocking the CPUs just to have better numbers in the reviews, that s why there can be 5% variation in some MT benches even when using the same CPU, Asus as i already pointed is one of the brand that has this habit whatever the CPU brand.

 

[Ranulf]

Author of DRAM calculator response to AMD release. Interesting if true.

https://twitter.com/x/status/1650791399699218433

"The volt problem has existed since the first Zen 4 reviews when the board duplicates VDIMM in CPU VDDIO MEM and VSOC. My colleague «destroyed» the CPU, but was asked to be silent about the problem at the time. AMD and ASUS policies have been unsatisfactory lately."

 

[Timmah!]

So what is the consensus here? If there is any?

Am i safe keeping my 7950x with DOCP on (have 2x32GB Gskill 6000MHz, not the EXPO version) or should i be better following that "Buildzoids Easy Guide for Hynix RAM OCing on AM5?" And specify those voltages manually based on that to values from that guide (or tad higer if it turned to be unstable) instead of leaving it on AUTO, as it currently probably is with just DOCP on? I presume i can improve performance that way, as a bonus, but that again it seemed to stable in regular use so far (did not do any significant stress testing), so if not needed, perhaps i would rather stick to what seems to be working just fine instead of meddling with it over some hypothetical, maybe unrelated issue.

EDIT: Still on older BIOS, 805, which was deemed to be one of the most stable one until maybe these most recent ones.

 

[DrMrLordX]

So what is the consensus here? If there is any?

Am i safe keeping my 7950x with DOCP on (have 2x32GB Gskill 6000MHz, not the EXPO version) or should i be better following that "Buildzoids Easy Guide for Hynix RAM OCing on AM5?" And specify those voltages manually based on that to values from that guide (or tad higer if it turned to be unstable) instead of leaving it on AUTO, as it currently probably is with just DOCP on? I presume i can improve performance that way, as a bonus, but that again it seemed to stable in regular use so far (did not do any significant stress testing), so if not needed, perhaps i would rather stick to what seems to be working just fine instead of meddling with it over some hypothetical, maybe unrelated issue.

EDIT: Still on older BIOS, 805, which was deemed to be one of the most stable one until maybe these most recent ones.

What's your current vSoC?

 

[igor_kavinski]

I had at some points smelled some "hot" smells when the system was going full tilt a few weeks ago but figured that was normal and wear and tear.

That's the CPU heatsink heating up due to the CPU boosting to 95 degrees C (within spec according to AMD). Lower the max temp down to 80 or 85 in the BIOS if it bothers you.

 

[DrMrLordX]

So I would really wait to see what goes on if its within your timeline.

Yeah I may do that. I was gonna finally tear down the 3900X this weekend but I can always hold off. At the very least, I would like to see what Gamer's Nexus presents from their study.

From what I can tell, it looks like vSoC values are going out of control due to stupid bugs that cause vSoC to match vDIMM which is kinda stupid. People that are hand-tuning their memory for vSoC of 1.2v and lower are likely not having this problem. But nothing is really certain yet.

 

[Timmah!]

What's your current vSoC?

No clue, probably set to Auto, if thats the default setting.
I played around it abit, enabled DOCP, enabled Dynamic OC Switcher, enabled Medium Boostit thing, enabled PBO Enhancement Lvl 3, and increased both PBO clocks to 5,85/5,75/5,55 and regular all core clocks to 5,25/5,15 per CCX and thats like it. Did not touch any voltages.

 

[DrMrLordX]

No clue, probably set to Auto, if thats the default setting.

You can use monitoring software to tell.

 

[utahraptor]

Hmm I guess I thought a bios from last week had fixed this, but I was wrong. There was a beta bios available two days ago, but it kind of had a disclaimer of "this might keep your CPU and mobo from dying, but it still does we warned you its beta so tough cookies." Thankfully the none beta is out now and I updated just now. Expo is still working at least at the default timing. I have not tuned them yet.

 

[CakeMonster]

I put VSOC on Auto, so it seems Asus has not quite managed to do what they said the explicit purpose of this BIOS was......


Edit: Oopsie, I can't read. The SOC limit is apparently just for the X3D CPU's....

 

[AdamK47]

Beta BIOS 1410 released for my Asus ROG X670E Crosshair Extreme. Updating now.

Includes AGESA 1007

ASUS ROG Crosshair X670E Overclocking & Discussion...

www.overclock.net

 

[CakeMonster]

Beta BIOS 1410 released for my Asus ROG X670E Crosshair Extreme. Updating now.

Includes AGESA 1007

ASUS ROG Crosshair X670E Overclocking & Discussion...

www.overclock.net

I noticed just after posting... this is fast moving.

VSOC limitation now working, even with DOCP:

 

[AdamK47]

My tweaked memory settings on the new BIOS. Set VSOC to Auto just to see what it would be set to. On previous BIOS versions auto would set VSOC to 1.35V.

 

[AdamK47]

I noticed just after posting... this is fast moving.

VSOC limitation now working, even with DOCP:

View attachment 80076

View attachment 80077

Unrelated: Have you tried your settings with GDM disabled?

Some settings you make in the BIOS may not be their true settings with GDM on.

 

[eek2121]

So im hearing its only on the X3D cpu's.
And its not just ASUS but all the vendors.

I am willing to guess it has to do with that boosting profile the Ryzen 7000 series has.
That mixed with the larger cache on the X3D could trip a massive power draw causing those burnout's as it does not have sufficient enough cooling.

https://www.reddit.com/r/Amd/comments/12xbgsm

Read the top comment on the reddit thread. The user is implying that Asus and Gigabyte BIOS automatically sets SOC and VDDIO voltages to 1.36-1.4 V, sometimes boosting to even 1.5 V in Windows, when EXPO is enabled. The regular chips can handle it. The X3D cannot.

The issue is SOC/VDDIO voltage. It was scaling up when EXPO or XMP(DOCP) is enabled. Nothing more.

Regular chips CAN be harmed by this, there are reports of non-X3D chips being damaged in the same way. It is pointless to allow SOC to rise above 1.3V. The UEFI currently does not cap SOC voltage on non-X3D chips. I recommend manually setting the voltages on non-X3D chips to the suggested values. Setting SOC to 1.3 should result in no stability issues.

 

[CakeMonster]

Unrelated: Have you tried your settings with GDM disabled?

Some settings you make in the BIOS may not be their true settings with GDM on.

Nope, just stayed with it because it was recommended with Z3. Something about odd numbers not working well regardless of GDM, so I just figured I'd keep it and use even numbers anyway. But it might be more complex on Z4 for all I know, I should probably look into it.

Its weird how you seemingly got a different VSOC value though, mine seem very much fixed at that value, and that is at the 'Auto' setting with DOCP Tweaked (it has varied before so I'm pretty sure it isn't controlled by DOCP on my sticks).

 

[aigomorla]

Yeah I may do that. I was gonna finally tear down the 3900X this weekend but I can always hold off. At the very least, I would like to see what Gamer's Nexus presents from their study.

From what I can tell, it looks like vSoC values are going out of control due to stupid bugs that cause vSoC to match vDIMM which is kinda stupid. People that are hand-tuning their memory for vSoC of 1.2v and lower are likely not having this problem. But nothing is really certain yet.

what im really curious is what is faster then.
The 7800X or the 7800X3D in gamings when there are limits in place on the 7800X3D.
If that is the case, the 7800X is most definitely a much better buy, as i do not see a physical fix which the cpu may need in pin layout to fix a issue of a burning socket for a cpu meant without limits.

 

[AdamK47]

Nope, just stayed with it because it was recommended with Z3. Something about odd numbers not working well regardless of GDM, so I just figured I'd keep it and use even numbers anyway. But it might be more complex on Z4 for all I know, I should probably look into it.

Its weird how you seemingly got a different VSOC value though, mine seem very much fixed at that value, and that is at the 'Auto' setting with DOCP Tweaked (it has varied before so I'm pretty sure it isn't controlled by DOCP on my sticks).

I don't select DOCP. I use Manual.

 

[Timmah!]

You can use monitoring software to tell.

I am on the older rig now, so cant do :-D
Will look into it over the weekend.

 

[Markfw]

You can use monitoring software to tell.

And if you are on linux ?

 

[mv2devnull]

And if you are on linux ?

With 'lm_sensors', if it has drivers for the hardware that is on the board. https://wiki.archlinux.org/title/lm_sensors

 

[Hitman928]

what im really curious is what is faster then.
The 7800X or the 7800X3D in gamings when there are limits in place on the 7800X3D.
If that is the case, the 7800X is most definitely a much better buy, as i do not see a physical fix which the cpu may need in pin layout to fix a issue of a burning socket for a cpu meant without limits.

This shouldn't effect performance at all.

 

[DrMrLordX]

And if you are on linux ?

Huh. Good question.

Based on a cursory search, this varies on a OEM-by-OEM basis. Asus, for example, is covered:

https://www.reddit.com/r/linuxhardware/comments/zjqikw/does_linux_support_sensors_on_new_am5_asus_x670e/

Not sure what granularity of data you get under Linux with something like hwmon or lm_sensors though. Personally I have not used it at all, and the redditor above was only using fan control.