newest BIO Z790 Hero Dark

[Shinchan0125]

I have update to the newest microcode.  BIOS 2001.  Prior version: 1801

Here is what i observe.  I have been ok with 1801

SVID Behavior [Best-Case Scenario] and CPU Load-line Calibration [Level 5]

HWMonitor when benchmark using R23 is around 280 Watt. VID Max during Bench mark lower down to 1.19.  R23 Score is around 41,000.  

After update.  It has become unstable.   R23 would crash half way through.  

I tried to 
Actual VRM Core Voltage [Offset Mode]
- Offset Mode Sign [+]
- CPU Core Voltage Offset [0.03500]

and did not work.  Still end up crashing the R23. 

End up by change Actual VRM Core Voltage back to AUTO.

and CPU Load-line Calibration [Level 6]

LL6 seem to stable it.

Now it seems stable.  But the during R23, wattage when up to 320W.  of course CPU temp at stress went up to 92C.  I know its normal.   for 14900K.  But before this BIOS update, at cold Room temp, it was only hitting 74C.

[weaver12]

It sounds like BIOS 2001’s voltage curve is more aggressive than 1801, which is why your 14900K is pulling ~320 W and spiking to 92 °C under R23. A few things to try:

ChatGPT said:

• Roll back to 1801 to confirm it’s a BIOS-induced behavior. If 1801 runs stable at lower volts/temps, then 2001’s microcode may need tweaking.
• In 2001, keep VRM Core Voltage on Auto and try a negative offset (–0.015 V or –0.020 V) instead of +0.035 V, since adding offset pushes the CPU to draw more power under load.
• If you want to stay on 2001, raising LLC to Level 6 or 7 is normal to prevent droop, but you can also fine‑tune a small V‑offset and drop LLC to Level 5 to get closer to previous wattage/temp.
• Double‑check that all power‑limit settings (Long/Short Duration Package Power Limits) are still set to stock values—2001 may have bumped those up. Setting PL1/PL2 back to the nominal 253 W/334 W (or lower) can cap the package power.
• Make sure your birthday calculator cooler is seated correctly post‑update; BIOS updates can sometimes kick fans into a different curve.

Good luck, and hopefully one of these adjustments gets your 14900K back to the ~41 k R23 range without the huge 320 W spike.

Regards,

[Shinchan0125]

When i was on 1801, i had .02 offset to keep it stable.    When it wasnt stable in 2001.   I keep adding offset to try out.. but that didnt work

next i try is Put to LL6.  Best case scenario,   Unlimited Power PL1/PL2 4095 (Unlimited Power).   It draw so much power.   

If you want to stay on 2001, raising LLC to Level 6 or 7 is normal to prevent droop, but you can also fine‑tune a small V‑offset and drop LLC to Level 5 to get closer to previous wattage/temp.    I am not able to get it back to LL5.  LL6 seem stable.

To keep temp low and stability, i change PL1 = 253, PL2 = 280.  This keep the temp around 80C during R23.  Score is around 39500.  

Do i need to keep system at high score?  No.  but its more of i had it before, but why i cant now.

[Vynra]

i really hate to be "that guy" but cinebench is super not a stability test. While it can be a good indicator, the best stress test is occt cpu, occt- linpack/prime95

[Shinchan0125]

found issue that contribute to instability.   I had been using TVB bios setting for almost a year.  TVB Boost until target +10.   By change this back to Auto.   Yes, i lost score to around 39500 in R23.  But it seems to be stable.  

One question i observe.  In HWMonitor.  I see Voltage VID.  during idle, i see vid seem to be changing and not stable.  i dont remember while idle, it keep changing.  In past, that would stay pretty much the same number.  But now.. they change between 1.395 - 1.405.

[Vynra]

This doesn’t make any sense. 

VID changes between frequency/load.

1.395-1.405 is not that big of a leap. Especially since VID and VCORE are temperature dependent. 

also your overclocking.