Having latest microcode from vendor is generally good idea. Now stop posting your instructions which are not recommended and endorsed by Intel or EVGA. 

 
Please chill out people. Bickering helps nobody.
 
 
 

Nereus, have you tried contact Intel about EVGA, if they communicated to them?

EVGATech_ChrisB
Thank you everyone for your patience as we are in contact with Intel, and we are looking into this further with our team. 
 
We will update everyone once more information becomes available.

Great! Thanks Chris.
 

kraade
I think the 13900 and 14900 should have had a label that they were intended for enthusiast only , so have you seen Vince's video lately ? 
https://youtu.be/b_NzKXOQWKk?si=TX1l7F5kV0pOkFDK
I think I know and I am late to this game . lol 

This is not an excuse. There is no dimension in which a necessary microcode update from the cpu manufacturer not being provided because the boards are 'enthusiast' is acceptable. The linked video also doesn't cover anything remotely related to the this thread topic. Why are you even posting it? From a business perspective, there are many people who bought the classified edition purely to support EVGA after they discontinued their NVIDIA partnership. The power draw issue is prominent in day to day usage. If a solution similar to competing manufacturers is not provided, EVGA can absolutely expect to not retain the business it saw with the launch of their z790 line. Their next MB launch will sell to a significantly reduced community, as those leaving look to other vendors who have the ability to keep up with microcode fixes. This is not hate. I love eVGA. I have been purchasing GPUs from eVGA exclusively since 1999. I waited for their z790 line to upgrade from my 8770k to 13900k specifically to support them as they dropped NVIDIA like an abusive spouse. The facts are that the classified consumers are looking for a rock solid platform to carry their slightly-more-than-mild overclocks through their system's gaming tenure. Nobody buying the classified cares about one-time or peak clock records. If it's not functional, it's garbage.

Right now, eVGA has a consumer base which
1. Is expecting stability AND longevity
2. knows the current microcode is unstable
3. knows there is an available solution to board manufacturers
4. has heard nothing at all from their board manufacturer
 
It should be expected for eVGA to not sell any of the next chipset line classified boards. Also, no one cares anymore about clock records. We're all doing 6ghz, and we're reaching diminishing returns on CPU frequency and performance returns in real life applications. Die shrhinking and power reduction is king going forward. Reaching 8ghz on z890 is going to be a non-story. It's not usable. It's not real, and no one is paying attention. If they can't sustain the same level of support that got us hooked, they can expect to lose us in our next build. There is no rebuttal to this. It doesn't matter how much you love eVGA. If you bought z790, you're currently up ****'s creak, and you have to decide if you want that kind of uncertainty going forward.

tcorrigan
*snip*
The power draw issue is prominent in day to day usage. If a solution similar to competing manufacturers is not provided, EVGA can absolutely expect to not retain the business it saw with the launch of their z790 line. Their next MB launch will sell to a significantly reduced community, as those leaving look to other vendors who have the ability to keep up with microcode fixes.
*snip*
Right now, eVGA has a consumer base which
1. Is expecting stability AND longevity
2. knows the current microcode is unstable
3. knows there is an available solution to board manufacturers
4. has heard nothing at all from their board manufacturer
*snip*
There is no rebuttal to this. It doesn't matter how much you love eVGA. If you bought z790, you're currently up ****'s creak, and you have to decide if you want that kind of uncertainty going forward.

EVGA appears to be winding up their business completely, so unfortunately it would seem the Z790 line is the last EVGA motherboards we will ever see.
 
A representative for EVGA posted yesterday that they are in contact with Intel regarding the microcode update and are looking into this further and will update everyone once more information becomes available. This seems to infer that we (hopefully) will see another BIOS issued for the microcode update.
 

Martin V
Like I said, lock all the cores first to the real speed of the cpu. If you have a 13900k it is 5.4 ghz, 13900ks/14900k is 5.6 ghz, and 14900ks is 5.8 ghz.

This will stop the boost algorithm from boosting the 2 cores so high, and then set the vcore to "adaptive" with a negative offset that can keep your cpu "vid max" voltage below 1.35v at all times.

Then set the vcore pwm frequency down to 400 and set the cpu power limit to 253 watts, and you are good to go.

Also, if you have Windows 11 not sure about Windows 10, but make sure you turn off "core isolation" first before going into the bios, and tuning your cpu.

Hi,

Would the settings for 13900KF would be same as 13900K?

These settings you mentioned can be done on Z790 Classified, correct?

Thanks.

Martin V
Yes they can be done on a Z690 and Z790 as the bios are the same. The Classified or Dark will both let you do these changes.

Thank you, I'll try this on my 13900KF on Z790 Classified.

Martin V
You got it! Also, a few other things too. Make sure you turn XMP "on", and for the voltage set it to 1.345v with a negative offset stafting at -75 and going up more if you need a higher negative offset.

You will probably have to play with the offset a bit and do your "cinebench" testing with "hardware monitor" at the same time to make sure the "max vid" voltage stays below 1.35v at all times.

For my 13900KS I set the voltage to 1.345v and a negative offset of -110 and my cpu never goes above 1.321v.

Hoped this helped.

Hi again,
So I did what you mentioned, limited all cores to 54, since I got i9-13900KF.

Set the voltage at 1.345 and offset value to -0.100

During Cinebench R23, for multicore test, using HWMonitor - IA: 1.275V, system agent: 1.331V, VID (Max) 1.242V, checking with EVGA's ELEET X1 tool, during the testing, 1.333-1.335V fluctuations.

For temperatures, P-Core (Max) - hit 93 Degrees Celsius, E-Core (Max) - hit 84 degrees.

Power Levels, PL1 & PL2 were consistent at 253.00 W

R23 - multi core score is: 33638 pts | single core score is: 2113 pts
MP ratio is 15.92 x

Is this acceptable? Any additional adjustments to be made?

That seems a bit like overkill to me, but I do have some pretty hefty cooling. The voltages that have been noted to cause damage are over 1.5 (people have seen 1.6 and even 1.7v in some cases lol). You should be completely fine at anything up to 1.4V, especially if you're boosting to higher clocks on single core loads. 5.8GHz single core boosts are usually completely fine at 1.4V too. At your current settings, you're throwing away about 15% of the performance a properly cooled 13900k should have in R23 multi-core. That will depend on your cooling and other settings though.

To be fair to the recommendations you're following, what that performance in R23 means to what you actually use your computer for may vary quite a bit. I noticed some improvements letting a few cores boost higher, but increasing all-core didn't help me much at all, so I limited all p-cores to 5.5 and put in some V/F offsets there to keep power down.

Martin V
Yeah that is not bad at all. You could try pushing the offset a little more by setting it to -105 to -115 and see if you can drop the temperature on the cpu a little more as well as the voltage too.

You should see an improvement in your "Cinebench R23" score too or at least the same with margin of error.

I use "Cinebench 2024" which scores the cpus way differently than "Cinebench R23" does. I know when I did my tests, I improved my score by like 20 points which isnt huge but still nice.

Also, make sure that you set the "vcore pwm frquency" in the bios from 500 to 400.

Overall though, not bad at all! Good Job!

Hi again,

I did set the Vcore PWM Freq to 400 from 500.

This time, I set the offset to -110 and my CBR23 score for multicore increased from 33,638 to 33,984.

However I noticed my clocks (Ps / Es) were maxing out at 4599 Mhz and 3599 MHz respectively.

My CPU temperatures were: P-Cores (max) : 91-92 degrees celsius , and E-Cores (max): 83-84. During the testing that is.

Not that I care much about the scoring points, just trying to set the sweet spot between performance gains and minimizing CPU degradation.

Thanks again, I'll continue to play with these variables and also check out CB2024.

Talon2020
 

EVGATech_ChrisB
Thank you everyone for your patience as we are in contact with Intel, and we are looking into this further with our team. 
 
We will update everyone once more information becomes available.

Thanks for the update, this MCU update needs to be available for ALL Z690/Z790 boards, no exceptions. 

I 100% agree, both platforms should get 0x129 MCU

My cpu must be made from the cpu gods themselves. 1.7 years on stock z690 dark kingpin bios settings. No issues yet. 

Sajin
My cpu must be made from the cpu gods themselves. 1.7 years on stock z690 dark kingpin bios settings. No issues yet. 

Same here with my 13700K. It's been OC'd slightly and in use for 2 years in November with no issue's.  I flashed the latest BIOS with the MCU (0.129) and with all the same BIOS settings, my package temp went up 2 -3C.  I flashed back to the previous and all is running back to normal.