Poor memory clocks caused by TG contact frame?

[nstgc]

I recently build a new system, but I can't get the RAM to run at XMP speeds without memtest spitting errors at me every 10 seconds. It's on the QVL, the power supply is was chosen from the Cult list (gold tier), and they're installed in slots 2 and 4 as instructed in the manual.

I first tried updating the BIOS, but that didn't help. This was followed by trying each DiMM individually, but the result is the same.

My only thought (which doesn't require I return anything) is that I might have botched the installation of the Thermal Grizzly contact frame. However, I'm not so sure this is the issue, since the RAM seems to run fine at Intel Stock of DDR5-5600. My understanding is that improper tightening results in losing channels all together, which is not what I'm seeing.

Before I start poking around, I'd like to get some opinions on this. I think it's the IMC, and that I need to RMA the CPU.

• PSU: 850W Super Flower Leadex Titanium.
• CPU: i7-14700k
• Motherboard: ASUS ROG Strix Z790-A Gaming WiFi II
• RAM: Team Group T-Create DDR5-7200 CL34 (CTCED548G7200HC34ADC01)

Aside from setting XMP, everything is running at stock.

 

[EarthDog]

Have you updated the bios to the latest? 7200 should be easy.

What cpu?

You can lose stability, period with a wonky install.

Also, not sure you can rma on the premise it won't overclock enough... if it runs the platform limit (5600) then technically it's working.

 

[nstgc]

Have you updated the bios to the latest? 7200 should be easy.

Yes, sorry, I should have mentioned that. Updating the BIOS was the *first* think I tried. It didn't help.

What cpu?

14700k

You can lose stability, period with a wonky install.

So, do you think loosening the screws on the contact frame might help? I'm trying to figure out something I can do, and the less I do the better I figure.

Also, not sure you can rma on the premise it won't overclock enough... if it runs the platform limit (5600) then technically it's working.

Yeah, "RMA" might not be the correct term, but I think I can return it. I got it from Newegg, and reading through their return policy, I think I can return it within 30 days for whatever reason, or, in the special case of the Holiday Season, before Jan 31st.


Not mentioned in my opening post, but everything else is at stock.

 

[EarthDog]

You can try loosening the frame, yes. That's where I'd start. Maybe check for bent pins just in case.

Update that sig when you have a chance.

 

[nstgc]

You can try loosening the frame, yes. That's where I'd start. Maybe check for bent pins just in case.

Okay, thanks. I contacted EK to see if the pre-applied thermal past on their AIO is the same as what's supplied in the tube. I don't want to clean up thermal paste, but I also don't want to mix different pastes, either, so I'll be waiting to hear from them.

I'll try loosening the frame first, then check the pins. I checked the pins with a loupe before putting the CPU in, and it seemed fine. If loosening the frame doesn't help, I'll check the pins.

If loosening the frame doesn't work and the pins are okay, would you guess it's the IMC?

Update that sig when you have a chance.

I will once I have a system that works. I'm sure most people would see DDR5-5600 and think nothing of it, but, well, I'm posting here, so clearly I'm not most people.

 

[nstgc]

You can try loosening the frame, yes. That's where I'd start. Maybe check for bent pins just in case.

Update that sig when you have a chance.

I've since done a lot of fiddling. What helped more was RMA'ing the CPU as that seemed like the next logical step. This CPU does do better, but it still can't pass MemTest86.

So I decided to try some other tests.

In particular I loaded up MPrime. It seems okay when run with the smallest test size (4k), at least for the short 10 minutes I attempted to start with, but when I bump it up, it fails. When the length is 288k (or larger I'd presume) one core will almost instantly fail. I'm not sure if that's normal for IMC/RAM issues or not, but it does seem like another data point.

Might this still be the motherboard? After I retest one DiMM at a time, the motherboard would be the next thing to return. (I sadly no longer have two RAM kits.)

The highest temperature I've seen so far was 91°C and that was for just a short bit before the AIO kicked up into high gear. (It's mostly been around 75°C.)

edit: Also FFT lengths of 200K seem to cause failures.

 

[Linux0s]

Just try loosening the frame. Please do it. Can't be too loose. The most u can, better will be.

Just enought pressure for keep ur CPU fix.

Don't forget ur CPU WaterBLock or A.I.O or Whatever dont need to be OverTight too. Just enought for giving a good contact on ur IHS or ur DIE if u have a Direct DIE.

***If u can replace the ILM *INTEL* for the Contact Frame, its so better so try Fix ur Problem my boy. Keep Going mate.

 

[nstgc]

I ended up RMA'ing that CPU. It was the IMC.

 

[Linux0s]

U can't said that if with the ILM thats worked before.

U said that before....

Are u really trying with the ILM before my man ?

 

[nstgc]

I'm not entirely sure of what you're saying, but I can assure you I am not "your man".

Regardless of what you are getting at, my memory issue was resolved by RMA'ing the CPU.