XMP, RMA or something else?

Associate
Joined
3 Feb 2009
Posts
921
Hi Folks,

So having never had any issues with XMP or memory controllers in general before, I rather naively went out and just bought a new cpu/mobo/memory and assumed that if the mobo manufacturer said they were compatible, that they would be.

I can get into the OS (didn't to a fresh install, that's what I'm trying next) and do quite a bit without issue, but in games (tried a few so far) at some point I'll get a hard lock requiring a reboot to get back. Having done some searching around I've seen that people had IMC issues with 12th gen and my specific board which lead me to try a few voltage tweaks to no avail, some only increased the speed at which the crash seems to be triggered.

The hardware is a 13600k, Asus strix z690-f wifi and 2x 16 Kingston fury beast 6000MT/s 36-38-40

Things I've tried:
'auto' memory control - letting the mobo handle default settings
XMP1
XMP2 -6000 and 5600 profiles
'tweaked XMP' - seems to be a mobo manufacturer enhancement to XMP settings

Things that work without issue:
memtest
prime blend test
general use (internet browser, video playback etc)

And the million dollar question is... is it a memory/compatibility issue, defective hardware or something I haven't thought of?
 
Last edited:
Didn't get anywhere near OC'ing as it hasn't been stable since first boot.
Tried clearing the CMOS multiple times at this point too, it should be running everything stock.

I've downclocked the memory to 5000MT/s but not as far as JEDEC so I guess I'll keep going to see if I can get it stable at all. Can't say I'm impressed with a kit that has compatibility claimed by the mobo manufacturer though.
 
I had a look at the QVl list and it should support your RAM at 6000 MT/s.

That was kind of my point :(.

Thanks for the advice both, I guess I'll be sending the memory back then.
 
The thing that I don't get is why memtest and prime95 ran without issue (prime ran a blend for 20 mins which is longer than any game I've tried lasts).

I haven't tried memory overclocking in a long time, is there something else I should be using to test stability?
 
To close the loop:

I spent another day or so trying various things including all manual timings, voltages etc and while I got a good few hours out of it without a crash at one point with the memory at 5600, eventually it gave up.
I did find that there were some 'asus ai OC' settings that were still on auto and seemed to make a difference to how quickly/frequently it would crash when I turned them off/to manual but ultimately nothing I could do would get it stable.

So the memory is on its way back and I'm ordering a different kit with some chips that aren't SK Hynix to see if that helps. I really hope it does as I don't want to have to send the CPU/mobo back too!
 
I had a similar problem with an Asus X99 system and it turned out to be the motherboard. Have you been monitoring temperatures, what GPU and PSU are you using? Did you try the RAM at 4800MHz JEDEC (SPD) settings?
The system is wc'd and CPU/chipset temps didn't get anywhere near concerning when I was running prime for 20+ mins. Maybe 80 odd on the CPU when it was boosting to 5.1ghz and no more than 40's on all the chipset reads.

I did try jedec, even at 1.35 volts with super loose timings but got the same results.

GPU is a 3080ti and psi is a kolink continuum 1200W.
 
I don't know whether this really is widespread or just affects a small minority, but some are claiming Asus has widespread memory issues with Zen 4:
Thanks, but I'm on Intel and also memtest was reporting no issues, hence why I was asking for something else to test with. I've seen plenty of reports of zen5 having issues with ddr5 but as far as I've found, Intel 13 seems to have a much better/more stable IMC for d5.
 
Sorry, force of habit to assume people have AMD CPUs. Could you try disabling C-States in the BIOS and see whether that makes the issue go away?
No worries, it's a fair assumption to make these days but Intel made more sense on price/performance for me right now.

I haven't tried messing with c states but I'm afraid it's not an option any more as the memory is with DPD on it's way back to ocuk :).

I've got a Corsair kit arriving tomorrow that's based on Samsung chips so I'm hoping that will just plug and play with no issues.
 
Whelp, it wasn't the memory :(.

The new corsair kit wouldn't even boot with both sticks in. Managed to get into windows with 1 stick but I'm down to CPU and/or mobo being the issue now.
 
So I've been away with work this week and haven't had much time to test, but I *think* I've cracked it and am absolutely kicking myself...

I have a PCIe gen 3 riser and hadn't set the bios to limit the gpu slot to gen 3. I tried a few more things before making that change and haven't seen a crash since, which puts me about 10 hours of use in at this point; well beyond anything I'd managed before.

It doesn't explain why the diagnostic led relating to memory was what it was stalling on during boot and why changes in dimm slots/number of sticks etc was having an impact on whether it would boot/how long before crashes... but if it stays stable then I'm going to keep using it and see how things go.
 
Yep, seems to be good so far. Now I just need to figure out what's causing this stuttering on my mouse input! Not ideal when you're playing a shooter and 1/5 clicks doesn't register...
 
Back
Top Bottom