Asus Z790-F XMP problems.

[phyxsius]

I've this issue going on since I built this PC. When I activate XMP profile. System goes bye bye. 

RAM installed on A2 and B2 as Asus suggests. Default XMP profile boosts up the RAM freq to 7800mhz. System boots ok. However, let alone gaming, even Chrome crashes constantly. Access Violation. Alternative XMP profiles result just the same.

Down clocking to 7600mhz delays crashing a little bit but not fully.
Down clocking to 7200mhz makes it more stable but after high loads result just the same. ( Video encoding, more than one game launched same time, CPU or GPU based rendering etc. )
Anything between 7200 and the default clock 4800mhz result boot loop.
default clock 4800mhz runs without any problems.

I've tried:
fine tuning timings and voltages. Bios updated to latest 31/03/2023. 
Remounting dimms after adjusting.
Different Slots.
Single dimm

I would really appreciate any suggestions on how to make it work so I can get what I've paid for.

Here are the gear that I use:
Windows 11
Asus ROG Strix Z790-F Gaming WiFi Intel Z790
Intel i9 13900k
G.Skill 32GB(2x16) Trident Z5 RGB 7800mhz CL36 DDR5 Ram (F5-7800J3646H16GX2-TZ5RK) 
Asus 1200W Rog Thor 1200W Platinum II 
ASUS GEFORCE RTX 4070 TI ROG-STRIX
ASUS ROG RYUO III 360
ASUS Rog Strix Gx601 Helios

[Silent_Scone]

Nobody removed comments, it was flagged by the system as spam (inadvertantly). As you have decided to respond again, I won't add it back into the thread.

You may not agree with it, however, it is the case. No vendor can guarantee overclocking. How do you propose one accommodates CPU silicon variance of each CPU? The maximum validation allows users to freely explore the overclocking potential. Naturally, when we run individual components outside of the default operating spec, some will have greater success than others. This has already been discussed in an earlier post which you freely agreed with. Your most recent post indicates you're still at odds with what fundamentally constitutes as an overclock. In this case, it is anything beyond 5600MT. When shooting for frequencies on the upper echelon of what is stable, often all one can do is suggest various tweaks available.

[phyxsius]

Yet Another thing you have taught me in this forum. Criticizing asus tech support is a spam. Thank you. I don't want you to get me wrong. After all it is not personal, It is about Asus. I'm not targeting you. Since you are the voice in this forum, I'm ought to do it through you. I'm sorry about that.

In Europe, we do things differently. You advertise what you can do. Not what you can hope or dream to achieve.
I've agreed everything you said about OCing. I still do. That doesn't mean that I need to accept false advertising. In my industry, when I publish about some add-on hardware compatibility with my base hardware, which is similar to what asus did - Tested and verified-. I'm obligated to see it through. If there is a problem, It is my responsibility to handle it.

A Qualified Vendor List (QVL) is a list supplied by a system or motherboard manufacturer of parts tested and confirmed by that manufacturer to work in their hardware. 

The keywords are: Qualified, Tested and Confirmed. I don't really care the truth about the CPU silicon varieties. I took my CPU and RAM kit to my friend who has APEX z790 MB on hand for test purposes. It worked just fine. Even handled 7800MT/s with such ease. Than I brought it back home with me to rule out any other possible hardware issues like power supply, GPU etc. Results was just perfect. Runs easily at 7600mt/s.  Even 7800mt/s.

Thanks to my dedicated friends they brought me some other i9 13th gen CPUs. 11 to be exact. Them all failed  plugged in to my MB but succeeded in apex z790. That rules out silicone variance in my case. If you still going to insist on CPU, I mean c'mon, the only "good silicon" landed on Asus devs but not me not even half decent withing 12 cpus. I'm not saying that you are wrong about silicon variety but sometimes it is just faulty manufacturing or steak is not cooked enough. 

I don't want to talk about support response to above mentioned results because of the "spam" thing. But here I am. Still no solution or acceptance of any possible faults from asus side.

[Silent_Scone]

---

@phyxsius wrote:

The keywords are: Qualified, Tested and Confirmed. I don't really care the truth about the CPU silicon varieties

---

Until such a time you do, we'll likely go around in circles 😉. Not caring for something isn't necessarily a valid argument for what is objectively the case. The Apex is a higher-end board geared towards maximum OC potential, sometimes at the expense of additional feature sets. It has a completely different memory topology. Please also be aware that memory vendors have their own QVL, perhaps you can ask them for some clarification on assurances also. F5-7800J3646H16GX2-TZ5RK is not actually present on the Strix-F Gaming QVL list, either - so it hasn't been validated by ASUS.

[phyxsius]

---

---

Until such a time you do, we'll likely go around in circles 😉. Not caring for something isn't necessarily a valid argument for what is objectively the case. 

---

It seems so, even though I've explained why it is highly irrelevant following my following comment on the matter.  If you keep going to ignore all that information. No point talking about it anymore.

This topic wasn't about F5-7800J3646H16GX2-TZ5RK anymore but F5-7600J3646G16GX2-TZ5RK. Which is in QVL.  I've indeed reached G.Skill about the issue also. They've told me that, they do test the products with fastest available CPU and MBs from the major manufacturers at the time of initial release. "Compatibility with any other MBs are handed over entirely to manufacturer to advance on. We are not good at updating our website with every other QVL list integration. If Asus claims that these kits are supported by their respected MB, you should have consult with them." Words might differ since I couldn't check my mails at this time but idea is there.

Since this is going in circles as you've pointed out. I won't be posting anything else here that might clear the situation for others that might have encountered similar problems. Unless something new comes up which will worth my time. I do thank you for your time and replies. Have a lovely day.

[Silent_Scone]

To clarify, did GSKILL guarantee overclocking in any sense? It doesn't seem they did from your reply.

Best of luck.

[phyxsius]

There was more than 15 emails went back and forward between us. I briefly summarized the content. But yes they did. They even suggested refund or replacement if I'm not happy with the overclocking results. I followed the suggestion and replaced with the same kit just to make sure it is not RAM related. Which was HUGE compared to what asus did. ( they couldn't even found the kit in the QVL until I've sent them clear instructions about how to find it ) I can share the emails there if it is not against the forum rules and also if they consent.

[Silent_Scone]

See here.

https://www.gskill.com/community/1584933243/1646978599/About-Memory-Overclocking#about-memory-overcl...

" PC memory is also controlled by the IMC (integrated memory controller) within the CPU, so the CPU IMC also needs to be strong enough to support the overclocked speed of the memory, and not all IMC of the same CPU model are equal in strength. So some extremely high speed memory kits are designed for extreme overclockers, which requires CPUs with strong IMC and overclock-oriented motherboards to be able to run at the rated specification. "

Please also refer to my post regarding memory topology and the impact of signalling on combinations of hardware.

Motherboard topology and variance can also play a part when talking about maximum validations when combined with a weaker CPU memory controller. 4 DIMM boards have greater signalling challenges to dial out things such as crosstalk, etc (including with only two modules installed). So comparing 8200MT on the Apex which is a 2 DIMM board is an apple-to-orange comparison. Moreover, different IC types will have different impedance values that may not favour one platform over the other. It's always best to opt for kits available on the motherboard QVL to ensure the best possible chance of compatibility.

Surely if the IMC is a factor, how can they guarantee an overclock? Of course, this is a rhetorical question. Because they can't [and do not], it's nonsensical to think any vendor can. I'm glad GSKILL have offered to replace or refund the kit for you 🙂

[phyxsius]

So I am a liar now? I'm beginning to lose respect. Like I do with Asus. They have indeed guaranteed that their kit will work at that speed. The rest is up to MB or CPU that can support. They even replaced the kit to prove the point. Of you were to work for Intel, I'm sure that you would be blaming Asus. This is what I see in your replies. Lack of objectivity. I'm telling you that I've tested with so many different CPUs and other stuff yet you still insist on CPU. Not even thinking about faulty board. Maybe this is a production specific issue, only affected my batch? You didn't even considered or suggested. Sorry but I'm done with this topic. 

[Silent_Scone]

I’m providing context and links from GSKILL. Perhaps where we went wrong was when you implied everything should work on the QVL which is a completely different issue to there being a board fault. This doesn’t change the consensus on overclocking.

Here is a previous sticky: https://rog-forum.asus.com/t5/z370-z390/understanding-and-using-xmp/td-p/797325

You’re free to initiate an RMA if you feel you exhausted all avenues; that’s a separate issue and you’re free to do so 🙂 👍

Feel free to try suggestions here if not already done so.

https://rog-forum.asus.com/t5/intel-700-600-series/windows11-crashes-and-the-pc-reboots-when-i-enabl...

[andino]

phyxsius, don't feel bad. I think a lot of people on Asus motherboards are having XMP issues right now. At least the ones that are trying anything over 7000. It seems like they had XMP working better in the earlier BIOSs but for some reason they aren't in the latest ones. That is totally on Asus though. Not you... 

You might want to look into manually timing and setting the voltages on them. I know a lot of people are getting great results over 7000 that way.