Z790-F Gaming Wifi I and Z790-F Gaming Wifi II wont go past 7000MHz on the RAM

[xtremefunky]

Hey guys, 

as many other people do, I like to play around with hardware.
And besides already having a binned and great Z790 Apex Encore System which runs rocksolid (VT3 and TM5) 32GB 8400C36 with a 14900KF I figured, 
I may build a 2nd PC just for the fun of memory overclocking.
I wasnt aware of the well spreaded info or rumors as "strix supports" would say that the board cant run past 7000MHz. I saw (as many others) 7800MHz in the QOL and was like: yeah. Looks like a decent board as you would expect from ASUS. 


Since I already had a 32GB Kit which wasnt that great, but still managed to produce 8400C40 I, I basically snacked off a cheap i9 14900kf which the owner said he wasnt able to get 8400MHz stable on his apex (with minmal effort) and a ASUS Z790-F Gaming Wifi I and Gaming Wifi II. 
I decided to go for the I and the II to see the quality difference in post and pre refresh. 

Full of enthusiasm I installed the i9, my own binned 8400C40 kit, combined with a 1300W Titanium Seasonic and tried to overclock the RAM, after installing the latest BIOS on both boards. 
Dont get me wrong. I definitely know how to OC DDR5. Setting up RTTs and also training algos is nothing uncommon for me.

I didnt do any special enduring stability tests, as I just wanted to hunt for the max postable and bootable frequency. 
Needles to say. Both boards did the exact same thing, besides ASUS advertising 200MHz more for the WIFI II. 
7200MHz was poststable with instant crash in VT3. 7400MHz postet on both boards and freezed after around 10 seconds in BIOS. Really EXACTLY THE SAME. 
(but hey, at least my little i3 12100 did the same on both boards and got 145 SP rating).

I was super disappointed. No words can describe what Im feeling since .. you could really say Im an ASUS Fanboy. 

FF. 1 week later. I was able to get my hands on 2nd hand Apex Encore for super cheap. 
Both the RAM and IMC of the i9 managed to get 8700MHz into windows and doing benchmarks. 
The joke is. This is even 100MHz better than my main system.
Withhin a few moments I was able to successfully run VT3 at 8400MHz.
(my little i3 still got 145SP and managed to run 7450MHz).

Besides the guys who can barely afford a Strix board and get brutally scammed, and cant back up the accusations of having a "bad IMC or bad RAM". I CAN. I did it for fun and I also got the money and results to back up to say: 
This board is not capable of running past 7000MHz even if you have golden IMC and RAM.
Be aware. Dont fall for this board.
The fact this CPU is able to post and boot and bench 8700MHz makes the whole story so much more disappointing. 

Dont get me wrong. I just want to show my experience with the boards I had. 
And how I feel about that. Maybe this helps other people to reconsider their buying choice.
There may be boards out there which can go past 7000MHz, but this shouldnt be the exception and definitely not involving any luck!

Greetings
-Chris

[Silent_Scone]

Hi @xtremefunky, Can I ask first what convinced you that having the same CPU on both a 2DPC board and a 1DCP board would ensure you obtain a similar frequency? It's important to temper expectations and to understand why the Apex exists in the first place.

I'm sure you realise, but more slots inherently introduce more signal crosstalk and impedance mismatches, reducing maximum stable frequencies. >8000 on the Apex doesn't ensure you over >7000 on 2DPC boards. However, you raise some valid points.

 Are you using a kit validated on the motherboard QVL?

Whilst the QVL doesn't offer assurances with overclocking, it does act as guidance. Using a kit validated at 8400 for use with the Apex, likely isn't the best fit for the scenario you're trying to build with the Strix, as binning on a 1DPC board like the Apex skews expectations. Auto rules scale with both latency and frequency. The memory vendor makes no assurances that stability will be found at a lower frequency with tighter memory subsets. It might be, but then it might not. If we consider you've mentioned using your own binned kit, this only amplifies this point - as the memory vendors instil a certain level of voltage guardband through a machine process. The trouble you have is working with those frequencies on 2DCP requires an equilibrium between parts that isn't quite as important when using boards such as the Apex, because the signal margin is that much tighter.

Does this mean using a 7200MT kit validated for the Strix will help?
Potentially, but not assuredly. Over > 7000MT on Z790 with more than 2 slots is certainly not within "easy" territory". 

To be clear, I'm not dismissing all of your concerns, but certain platforms have limitations. I feel where there's a valid point is just how aggressive some validation is and how likely one person to the next can achieve things. These are valid concerns I have put forward recently. However, this is overclocking, and there will always be an element of luck - especially when approaching frequencies on the upper echelon of the validation table. If it's on the MB QVL, the kit would have been tested as stable using the same standard of testing as the memory vendors use.