Maximus XI Hero and G.Skill 4133 Mhz Cl19 booting problem

[Ch1re]

Hello,

I have a problem with: Maximus XI Hero (Wi-Fi) + 9900k + G.Skill 4133 Mhz Cl19 (F4-4133C19D-16GTZR - listed on the QVL) combo.
With enabled XMP, it's almost impossible to pass the booting process - 9 out of 10 attempts ends with POST code 23.
However, in rare cases when it has passed booting process computer was 100% stable - also booting problem doesn't exist when I disable one of the RAM channel by BIOS which makes that RAM working in single channel mode.

It seems that the memory training is not able to handle that RAM even though it is listed on the QVL.

BIOS: 0602

[Silent_Scone]

Aurosonic wrote:
I'll wait till APEX XI come and test my CPU with new memory kit, so it will be evident if it was motherboard or cpu imc issue. Then i'm gonna change my kit to 2x16 4000CL19 and will OC it to 4133CL17. One of my friends got this kit on Gene XI and it works perfectly.

Two DIMMs are a lot easier at the electrical level both on the CPU and the board. Obviously, it's not possible to test 4 modules at a given speed on the Apex :D.

If I can source your kit, I'll post some results with the Formula XI.

vinceX wrote:
Makes sense. The kit I have is from the same batch and has all consecutive SN. The main reason I got 2 pair of a 32GB Kit is that I cannot find any G.Skill 4000mhz CL19 RGB Kit for 64GB.

Yes, binning that many modules to work at that frequency with enough guardband is tough. Which is why you won't find many/any kits.

[Aurosonic]

Silent Scone@ASUS wrote:
Two DIMMs are a lot easier at the electrical level both on the CPU and the board. Obviously, it's not possible to test 4 modules at a given speed on the Apex :D.

If I can source your kit, I'll post some results with the Formula XI.

Yes, binning that many modules to work at frequency with enough guardband is tough. Which is why you won't find many.

Yes, i know about 2 slots APEX. I'll be fine with 2x16 on APEX instead of 4x8 on Formula at same frequencies. Just hope it's not my IMC issues.

Here's my results on Formula: https://imgur.com/a/eR4CWON
Worked for 3 months till 1 of the sticks has died.

[Aurosonic]

As i expected, issue was my motherboard. Switched to Apex XI and so far so good.

[Silent_Scone]

Aurosonic wrote:
A got some memory issues with my Formula XI as well. Memory kit G.Skill F4-4266C17Q-32GTZR
Wasn't able to pass stresstests at 4266CL17 no matter which vccio and vccsa, tried up to 1.4v
Been able to pass tests by downclocking it to 4200. After 3 month of successfull usage at one day my system hang so even reset didn't work.
I turned off the system and on next boot i got postcode CC. None CMOS reset, retry, memok didn't help to pass this CC postcode.
Then i've tried to check each memory stick separately and found that 1 of the sticks caused the problem. I've tried it in each of four slots and it gave me postcode CC every time.
Once i remove this stick and leave three others, system boots perfectly fine but i had to downclock it to 4100 to make it work stable (seems like cuz of T-Topology).
Tomorrow i'm gonna sent memory back by warranty.

Got another kit, 2x8 this time F4-4600C18D-16GTZR. Same as previous kit it works stable only at 4100CL17 due to T-Topology. Now i've ordered APEX XI Motherboard and hopefully it'll solve all my issues and i could run at least 4400CL17 stable with my F4-4600C18D-16GTZR kit.

Aurosonic wrote:
As i expected, issue was my motherboard. Switched to Apex XI and so far so good.

Those speeds can be tough on 4 DIMM boards, you need a good CPU and, if needed, good methodology when tuning in IO. Of course, even then there are no guarantees of achieving the kits validated speed and timings.

Took a while, but I sourced that kit.

F4-4266C17Q-32GTZR on Formula XI @ XMP (1 Hour of Stress App Test)

[EtoEgor]

Silent Scone@ASUS wrote:
Those speeds can be tough on 4 DIMM boards, you need a good CPU and, if needed, good methodology when tuning in IO. Of course, even then there are no guarantees of achieving the kits validated speed and timings.

Took a while, but I sourced that kit.

F4-4266C17Q-32GTZR on Formula XI @ XMP (1 Hour of Stress App Test)

Please help

[EtoEgor]

Exactly same problem with HERO XI BIOS 0805, and HyperX Predator 3200Mhz 32GB (8x4) HX432C16PB3K4/32 (Hynix) + 9900k. Sometimes the system may not start. I tried different frequencies and voltages - full random. The system can start ten times in a row, but not start the 11th time! And it will not start until you reset the BIOS settings or enable MemOK (sometimes memok suggests 1.45v and 1.4 vccio&sa for xmp 3200mhz omg!!!) At the same time, I calmly pass the 24 hour stress memory test if I can boot even with OC ram to 3466 16-18-36 and 1.1 vccio&sa!!

Please help!! On my old z370 g-gaming everything was fine for 6 months! 0 problems! I think this is bios problem!!

[Silent_Scone]

Under the DRAM timing page, enable MRC Fast Boot and soak test the system. You might want to consider using a different method of testing memory stability (such as HCI Memtest or Ramtest)

[EtoEgor]

Silent Scone@ASUS wrote:
(such as HCI Memtest or Ramtest)

I use exactly these tests 🙂 and can pass easily them for 24h

So I don't understand why I cant pass on boot test sometimes...*

[Silent_Scone]

EtoEgor wrote:
I use exactly these tests 🙂 and can pass easily them for 24h

So I don't understand why I cant pass on boot test sometimes...*

The difference between what constitutes as electrically valid and what is valid in the OS can some times differ enough to cause these things. You can possibly dial this out by adjusting VCCSA/VCCIO or alternatively using the VCCSA/VCCIO boot voltage.

[EtoEgor]

Silent Scone@ASUS wrote:
alternatively using the VCCSA/VCCIO boot voltage.

How?

And what if I leave everything as it is? The system is completely stable if MRC Fast Boot is enabled...

I still think that your bios is bugged.