Z790 Maximus Extreme + 7200MT/s DDR5 Instability

[Selder]

I am experiencing stability issues when using the F5-7200J3445G16GX2-TZ5RK sticks on an Asus Z790 Maximus Extreme when running the DIMMs on their XMP I or II profile. These instabilities can be any of the following:

- random PC reboots when idle or stressed
- WHEA event code 17 in Windows event viewer
- PCIe device and/or PCIe bus errors

When running the F5-7200J3445G16GX2-TZ5RK sticks without the XMP I or II profile, the rig boots but obviously at default clockspeeds for the RAM.

Here are the specifications of my rig (except for the 2nd kit of DIMMs, currently running only 2 DIMMs in slot A2 and B2) : https://pcpartpicker.com/list/MjQJW4 . BIOS and MEI have been updated to their latest version. Windows 11 22H2 with a fresh (yesterday) Media Creation, standard installation no GPO's or anything. Running without Antivirus, and Memory Isolation is disabled (required for Intel XTU Tool).

I am currently running the same rig with the F5-6000J4040F16G (slower and looser timings) and that's perfectly stable (I can torture the rig with Techpowerup's Memtest64 without error) and run any benchmark and get consistant results. Performance is awesome 🙂

I'm wondering if anyone has an idea what the issue could be. As far as I'm concerned, this should all work "nearly out of the box" (aka: set XMP and done) and I shouldn't have to fiddle around with the gazillion memory tweaking options, should I? 🙂

[Shinchan0125]

question? How do you set vdd and vddq to 1.45? ON my Hero board, when i put in value 1.45, it change to 1.435?

[Hopper64]

Shinchan0125 wrote:
question? How do you set vdd and vddq to 1.45? ON my Hero board, when i put in value 1.45, it change to 1.435?

There should be a setting just above the DRAM voltage that says something like High DRAM voltage auto/enable/disable. You have to enable that if it is supported on your board. Not in front of my machine at the moment. I will post a pic if needed. Â*

[Silent_Scone]

As above, this setting toggles a setting on SPD that allows for higher than 1.43v on certain memory IC. It can be found above VDD and VDDQ voltage options on the Extreme Tweaker menu.

[Shinchan0125]

i have g.skill 7200. I was able to oc it to 7600 with CL36, 1.4 volt. No error at all on mem86. However, during AIDA64 Memory stress test. After 5 min. My 3 monitor start to freak out. random loss video... monitor flicks.. on and off and couldnt get a image... I have to use ALT+f4 to terminate Memory Stress Test.
Back it down the 7400 CL34 and AIDA Stress Test. No more random screen off. that was weird. mem86 didnt have issue.. but Stress Test it bring monitor/graphic issue.

[Selder]

I received some manual settings (which really are the XMP values like tCL-tRCD-tRP-tRAS 34-45-45-108) from G.Skill support.
Configured as 6800MT/s it passes a memtest86, but a memtest64 crashed in Windows afterwards.
7200MT/s settings gave hundreds of memtest86 errors in the first minute.

I have been in contact with Asus support too, but it's difficult to get around the 1st helpdesk level (who wanted to explain what O.C. was, and then nearly closed te support ticket with "it's an overclock, we don't guarantee anything".

Let's see what they both say in the coming days, might need to RMA both.

[xeromist]

Selder wrote:
I have been in contact with Asus support too, but it's difficult to get around the 1st helpdesk level (who wanted to explain what O.C. was, and then nearly closed te support ticket with "it's an overclock, we don't guarantee anything".

Technically, they are correct. XMP is actually an overclock, but being on the QVL means someone has tested it so it should be an achievable overclock. Unfortunately if XMP is failing you I seriously doubt any level of technical support is going to be able to assist. If your hardware is failing a validated baseline then it just becomes a matter of hand tuning and testing like you've been doing.

Also keep in mind that the memory controller is on the CPU, so even if you swap the motherboard and memory kit you could still have issues. A particularly poor CPU is just going to struggle with any faster kit. That said, it's probably simplest to swap the memory first to see if you can get a better kit, then start swapping other hardware.

[Selder]

Or ... just wait until Asus fixed these kind of things in their bios 🙂

Beta 0810 fixed all issues for me.

Running the 7200MT/s kit with XMP_Tweaked profile, enabled all Asus/AI/Turbo/Jet-eigine functionality and I'm scoring the highest benchmarks now.

YEAH!

Rig specs: https://pcpartpicker.com/list/MjQJW4
3DMark bench: https://www.3dmark.com/spy/34326243
Userbenchmark result: https://www.userbenchmark.com/UserRun/58068362
Screenshots BIOS:https://imgur.com/gallery/SYrVDyS

Thanks everyone!

[Selder]

And apologis, with all due respect - but I don't think it's too much to ask for this combination to work when they advertise with 7800(O.C.°+ in their manual. Note the plus character. 😀Â*

This crap should just work, nothing more nothing less. Â*You can't have it both ways, advertising for extreme and then when you want to go extreme the answer is "no not that extreme" haha 😀 Â*

[Silent_Scone]

Selder wrote:


This crap should just work, nothing more nothing less.

This kind of mindset is akin to asking why certain individuals are able to perform physically exerting tasks and why others can't. It's objectively incorrect.

Please read this sticky:
https://rog.asus.com/forum/showthread.php?98041-Understanding-and-using-XMP&p=690748&langid=5



It's impossible to account for every CPU in auto rules outside of reference code, nor is it possible to make assurances on processor IMC variance. The higher the frequency on the validation table, the more likely you will need manual input. So despite being stone walled by support - what they are saying is correct.

[Selder]

We could spend a whole day making comparison after comparison, but all these comparisons will ignore the simple fact that an end-user should expect the products he buys to meet the manufacturer's expectations, right? You simply cannot have it both ways: praising your product to the sky and luring customers with "O.C." (even with "+" symbols) and then saying afterwards "oh ehhm well, ehh, it doesn't work like that" 😀Â*

This obviously doesn't go for Asus only: G.Skill, Intel, ... they all do the same.

And of course nothing is guaranteed, but that all the components I chose do now all work well together as advertised, that is the most important thing and that is what I was talking about. I didn't expect anything else, Asus always delivers 😛