MSI Z890 Tomahawk - Inconsistent VCore starting with BIOS version A73 and later

P

patsc158102e1

New member
Joined
Feb 8, 2022
Messages
26
Hello,

I have been experiencing a problem since BIOS versions A73 and later, where the set voltage values are not always retained after powering off and on. Since BIOS A73, the voltage after startup is either increased or decreased for some voltages. In any case, the configured values are not always applied at each startup. If everything is set to auto, you might not notice this or it might not occur. However, if BIOS settings have been changed, it’s possible that altered voltage values are present when the PC is turned on or after a reset.

You can test this yourself by setting an offset for the VCore and performing multiple resets, then always entering the BIOS and checking the VCore value. After 2-3 resets, the VCore usually differs. For example, I have an offset of -0.065V, which results in a VCore of 1.136V. But after a reset or a fresh power-on, the VCore might be 1.108V, which can cause crashes for me. The A72 maintains a steady 1.136V VCore, but all later versions do not. Sometimes, after powering on or resetting, the voltage values change.

I would like to know if I am the only one experiencing this or if other users have observed this behavior as well.

Thank you very much.
 
Sort by date Sort by votes
Well, do you see voltage difference after couple restart in BIOS? But if the beta was posted in Aug and latest BIOS for Carbon is from July, I expect a more recent BIOS for Carbon.
 
Upvote 0 Downvote
plutomate said:
Well, do you see voltage difference after couple restart in BIOS? But if the beta was posted in Aug and latest BIOS for Carbon is from July, I expect a more recent BIOS for Carbon.
Click to expand...
I haven't set a offset. One of these days I'll test this. My issues would be instability just by entering bios a few times and saving. To get stable I had to reset cmos and load last stable settings. After a few more bios loads. Things would get unstable again. really sounds like same issue. thanks
 
Last edited:
Upvote 0 Downvote
Unfortunately, the new BIOS 7E32v1AA0 for the Z890 Tomahawk is a hard setback. The inconsistent VCore behavior is back. Also, the V/F curve of the Ring Voltage is not carried over again, and Undervolting Protection is now enabled by default in the BIOS. So you should disable it beforehand if you want to undervolt. It also seems that the same offsets from my Beta 1A94 still do not lower the voltages in 1AA0 as desired. But that may also be related to the generally inconsistent voltage behavior.

I didn’t spend much time with the 7E32v1AA0 BIOS anyway, because I simply don’t trust the new BIOS and am fed up with constantly finding new settings to work around the BIOS bugs. Therefore I backed again to Beta 1A94. More precisely, to my own modded 1A94, where I have the latest microcode 1D (Yes, MSI you are always too slow and only have 1B in the 'latest' BIOS) and even the latest Management Firmware 2095 installed (And yes, MSI you only have the 'old' 2018 firmware in the 'latest' BIOS). If someone wants the latest Management Firmware, at least this is very easy to do by googling for 'MoKiChU Intel ME Z890'.

All in all, a disappointing BIOS release for anyone undervolting on the Z890 Tomahawk. I hoped the new BIOS would build on the Betas -1A92, 1A94, 1A94-, but it seems to be built more on the broken 1A90, the official predecessor. If MSI follows a kit-based approach, I don’t see good prospects for other MSI Z890 boards either.

This is just my experience on my PC and personal opinion, and is only intended to serve as information.
 
Upvote 0 Downvote
Thanks, but I only wanted to share my experience and not discourage anyone from trying it themselves. Every PC configuration is different.
It should rather be a kind of heads-up about potential issues to expect, and that everything must be re-tuned and tested if you flash the 1AA0 BIOS.
An existing old setting only provides the rough direction at best, but a 1-to-1 implementation is something one can only dream of.

I'm disappointed, because the betas, aside from a few quirks, were very promising, and now suddenly this happens. I rub my eyes and say what the hell...:angryfire:
Optimists ahead!:beta: Who dares?
 
Upvote 0 Downvote
I just updated to latest BIOS and don't see any new issues.
All stock settings apart from enabling XMP, slightly reduced VCC Core, CPU Core, CPU Ring (about 0.070-0,100v), VCC SA 1.080v

VCC SA applied voltage is shown now (maybe was shown in Beta as well?)
It was set to 1.200v, I reduced to 1.080. Confused this with SOC SA... SOC SA on Auto

All running good so far.
 
Last edited:
Upvote 0 Downvote
Nice to hear that it works for you and that you have a stable setup you’re satisfied with and that you’re having fun. Don’t worry about things like Undervolting Protection. It’s only there to prevent certain limits from being undershot. I don’t know your setup, but I suspect you’re being very conservative. And as you yourself have written, only a few values were changed. I consider it unlikely that problems will arise.

I undervolt on a larger scale. Almost all voltages are touched and changed for me. I deliberately push to the point where the system becomes unstable. Then a safety margin is added and after various tests again, once the system has proven to be stable, it will be put into productive use.

But to do this, it’s necessary to disable Undervolting Protection, because Intel often sets the lower voltage points very restrictively, since you have to assume the worst CPU to ensure broad compatibility across all systems. Your CPU or mine, however, might operate with less voltage than the one set by Intel due to better binning. That requires a lot of testing and patience, at least at the beginning, until you learn to estimate your CPU/MB/RAM combination.

I already said elsewhere that normal users or light UV should not cause problems. But you’re potentially sacrificing a bit of efficiency and not using the full potential for power savings.

But hey, anything is possible and nothing is mandatory. That’s the nice thing. Everyone can run their system the way they want and have fun with it.
 

Attachments

  • MSI_CB_R23_BENCH_UV.jpg
    MSI_CB_R23_BENCH_UV.jpg
    294.2 KB · Views: 38
Upvote 0 Downvote
I personally would like to always use the latest BIOS build. Unfortunately, 1AA0 is unusable for my setup, as it continues the fluctuating VCore from earlier official BIOS builds. I always start with a clean BIOS and reset everything beforehand. After the BIOS flash, everything is reconfigured again.

Yet, the 1AA0 BIOS on my Z890 Tomahawk shows this inconsistent voltage behavior. This is not the case with the betas. I can’t say more about it.

Therefore I still use 1A94. But that’s a decision everyone has to make for themselves and try. The profile of the current BIOS is quickly backed up, so you can flash back quickly if needed.

If the question isn’t: Which BIOS should I use, but which BIOS was used in the CB_Bench_Picture?! It is 1A94.
 
Upvote 0 Downvote
That’s a concerning issue unstable VCore after BIOS updates can definitely explain random crashes, especially with offsets applied. Hopefully MSI addresses this soon; until then, rolling back seems safest… signing off like this after late troubleshooting feels like good night shayari Hindi for PC builders.
 
Upvote 0 Downvote
I had a system freeze (Windows 11, mouse, keyboard, clock - all stuck) with 1AA0 exactly when plugging in (or out?) a network cable.
Nothing apart from "unexpected power loss" in Windows event logs, so it was totally locked up.
Not sure if related but this was the first freeze since I use this system.
 
Upvote 0 Downvote
gert.riegle124a0299 said:
I had a system freeze (Windows 11, mouse, keyboard, clock - all stuck) with 1AA0 exactly when plugging in (or out?) a network cable.
Nothing apart from "unexpected power loss" in Windows event logs, so it was totally locked up.
Not sure if related but this was the first freeze since I use this system.
Click to expand...
that seems more like a driver prob. but who knows.
 
Upvote 0 Downvote
I think it's a bug in this versions 7E32v1AA0, E7E32IMSI.1A93, E7E32IMSI.1A90 after upgrade the PC will set on wake-up setting via bios, USB keyboard and mouse occasionally turn on by themselves without touching anything it starts. (MAG Z890 TOMAHAWK WIFI)

What happens after 2026, Z890 motherboards will no longer have bios update support ? (Is coming NOVALAKE S with LGA 1954 socket and Intel 900 series chipset.)
 
Last edited:
Upvote 0 Downvote
gert.riegle124a0299 said:
Maybe -0.070 on E-Cores was too much now with 7E32v1AA0
Click to expand...
My weakest P-core would intermittently freeze/hard-lock my PC during Prime95 Small FFTs when undervolted just beyond the edge of stability with the A92 beta BIOS. Adding a mere 5mV ostensibly fixed it.

If AA0 exhibits inconsistent per-core voltages across reboots like A90 and its predecessors, a -70mV undervolt that is right on the edge of stability with A94 or an earlier A9x beta BIOS might cross the threshold to instability with AA0 due to an intermittently lower base voltage against which the offset is applied.

I am sticking with A92 for this reason. Maybe I'll flash to A94 at some point, but I will not use any official BIOS with unreliable voltage fluctuations, such as AA0. My undervolt has fairly narrow guardbands, so intermittent ~20mV fluctuations in base voltage due to a BIOS bug would be very unwelcome. Pity AA0 couldn't retain the fixes from the A9x beta BIOSes. We were making such good progress with those...

Can anyone confirm whether the iGPU XMP hardware acceleration workaround from A93 has also regressed in AA0?
 
Upvote 0 Downvote
Back
Top