Overclocking RAM to 6000mhz on 7800x3d instable

[EarthDog]

It seems, there is a problem somewhere else, am I right?

I'd guess so. The event viewer doesn't share any details?

 

[Kiso133]

I think event viewer tells nothing helpful:
5 times id 201 mentioned above
and
1 time id 2 ("Session "Microsoft.Windows.WindowsUpdate.RUXIMLog" failed to start with the following error: 0xC0000035")

But as I said: im not sure, whether the restarts occur directly after this error messages or a while afterwards...

Does anyone have any ideas on what component it can be? Maybe it's no hardware issue at all?

 

[67Elco]

I'll assume there was no resolution to this problem. Another note would be to raise the LLC...spikes at lower voltages often result in crashes and you will receive no useful information as to the cause(s). DDR5 is very sensitive in many areas, unlike DDR4. I'm of the mind to ignore the EXPO function of the motherboard and do these things manually. Motherboard manufacturers are just guessing and there are just too many variables concerning memory and cpu's to give a one setting covers all solution.

 

[EarthDog]

Motherboard manufacturers are just guessing

In what way? EXPO, like XMP, is read from the DIMMs. There isnt any guessing involved. Mobo mfg test against those settings to determine if its working. Unless you get to really high speeds or high capacities, an overwhelming majority are plug and play... especially at reasonable speeds.

 

[67Elco]

In what way? EXPO, like XMP, is read from the DIMMs. There isnt any guessing involved. Mobo mfg test against those settings to determine if its working. Unless you get to really high speeds or high capacities, an overwhelming majority are plug and play... especially at reasonable speeds.

Motherboard settings such as EXPO and XMP lock you in to their final solution which is clearly not 100% effective as evidenced by the above posts and thousands more just like them. DDR5 has a very narrow window of success baked in...DDR4 does not have these liabilities.

There is nothing really wrong with DDR5 memory...it's just the nuances it requires is greater than solutions readily (read easily) available. Yes, a large percentage will operate as advertised even within the constraints of motherboard instructions (read limitations).

 

[EarthDog]

There are hundreds of thousands unposted successes too.

AMD has histoircally been more finicky than Intel. But at reasonable speeds, ddr5 is still plug and play 95 times out of 100. There will always be exceptions and problems, and ddr4 is quite a bit more mature which helps that cause... though there's a few more differences (pmic) this gen.

 

[67Elco]

This kit is great out of the box, but if you take the time to manually tune it is outstanding for the price. https://www.newegg.com/g-skill-32gb-288-pin-ddr5-sdram/p/N82E16820374369?Item=N82E16820374369

 

[Woomack]

I'm almost sure that right now, it's the same stuff as Trident Z5 and you can push it above 8000 ... but with different heatsinks and cheaper. From some brands like G.Skill, Kingston, or V-Color, I see only Hynix A-die in 16GB modules and M-die in 24/48GB modules, so it can't get any better. Corsair keeps releasing weird kits to convince people with lower prices, and you can find Micron-based 6400-6800 CL40/42. Most other brands have almost only Hynix too, but it's easy to check as 6000+ CL30-34 is only Hynix.
Samsung is not existing in DDR5 right now. The first IC was pretty average, and no one is using their ICs in gaming/enthusiast kits anymore.