Ryzen 7800X3D + Corsair Dominator Platinum RGB 32GB - Issues

Associate
Joined
25 Aug 2024
Posts
6
Location
ES
Hello,

First of all, thank you to everyone who tries to help.

Here's the thing, I've decided to upgrade my old Ryzen 1600 to a 7800X3D:

CPU: 7800X3D
Mobo: ASUS TUF GAMING B650M-PLUS WIFI
RAMs: Corsair Dominator Platinum RGB - CMT32GX5M2D6000Z36

The issues is that, from the very first time I tried to use the EXPO profiles, the computer took ages to boot up (MOBO DRAM LED on), approximately >1 minute to POST, and although at some point I managed to get past POST successfully, all it takes is a reboot or shutdown to fail POST the next time. In short, I can't boot the RAMs at anything above 5200MHz, and even when I can, the POST takes >1 minute and fails most of the time (at 5200 seems to work fine, and POST times are around 10/15 sec).

I've been trying everything, watched lots of videos about this, tested profiles I've seen online for RAMs with similar modules, but to no avail. I no longer know if the problem is with the mobo / cpu, or with the RAMs because of point 2 (below).

To make things easier, I'll summarize some of the tests I've done:

1 - Test each DIMM individually (1x16) only in the first slot (according to the manual - DIMM_A2) at 6000MHz. Result = POSTs with both DIMMs at 6000MHz, but takes >1 min.
2 - Test each DIMM individually only in the second slot (this is not the correct layout in the manual, but I wanted to test if the mobo slot was damaged?), at 6000 MHz. Result = POST failure with both DIMMs.
3 - Reinstall the 2 RAMs in dual channel, according to the correct position, and tested setting the voltage to 1.4v for the RAMs (VDD and VDDQ) @ 6000 MHz. Result = Usually it POSTs the first time (>1 min POST time), but fails the next reboot or after 2 reboots.

Finally, I tested other set of RAMs from a friend (vengeance 32GB - CMK32GX5M2B6000Z30) and it did POST at 6000 MHz first try, then I tried 3 reboots and the machine always rebooted without failing to POST and, a curious thing, is that the time to POST was much shorter (15 seconds +/-). Maybe it was just luck, and maybe it would fail after a few more reboots, but just the fact that the POST time has normalized makes the behavior very different.

In short: it seems to be the RAM, right? The point is that when I tested booting them individually, they only booted at 6000 in the first slot, and not in the second (DIMM_B2), which made me wonder if it was something on the board...

However, when I tested with these other RAMs, coincidence or not, the system seemed much more stable at POST, and the POST times were MUCH shorter.

Any ideas other than changing the RAMs?
 
Finally, I tested other set of RAMs from a friend (vengeance 32GB - CMK32GX5M2B6000Z30) and it did POST at 6000 MHz first try, then I tried 3 reboots and the machine always rebooted without failing to POST and, a curious thing, is that the time to POST was much shorter (15 seconds +/-). Maybe it was just luck, and maybe it would fail after a few more reboots, but just the fact that the POST time has normalized makes the behavior very different.

RAMs: Corsair Dominator Platinum RGB - CMT32GX5M2D6000Z36
Do you know what ICs those kits use?

From what I've read on another forum, the C36 kits are more likely to use Micron and some AM5 boards seem to be dissatisfied with this combination and struggle to train/optimise the board for them. Whereas, the C30 kits are more widely supported and the quoted kit (CMK32GX5M2B6000Z30) is present on the QVL of your motherboard.
 
No idea unfortunately.

The funny thing is that when I manage to boot up at 6000 or 5800, at first glance the system seems minimally stable, but it's always at POST that the problem arises, oh, and the POST times are always exaggerated.

My fear is that it's a problem with the mobo/cpu, but the truth is that with the C30 kit the POST times at 6000 were normal, completely different behavior. I also ran cinebench with the C30 kit without crashes (I know that's not enough for a stability test).
 
Last edited:
Not that this will help you but this kit
CMK32GX5M2B6000Z30 runs fine at 6000 on Asus board, checked few motherboards before purchased to make sure is on QVL.
I'm sure I have seen people adjusting figures manually is well if EXPO not working as it should, maybe worth a shot.
 
Out of interest does your BIOS have this option - ASUS Enhanced Memory Profile (AEMP)

Seems to be a way to run RAM faster without the EXPO being available. Maybe a temporary boost while you consider your options.

EDIUT: Tried on my B650E and no AEMP. More reading and it only appears on DIMMS without an XMP/EXPO profile.
 
Last edited:
Not that this will help you but this kit
CMK32GX5M2B6000Z30 runs fine at 6000 on Asus board, checked few motherboards before purchased to make sure is on QVL.
I'm sure I have seen people adjusting figures manually is well if EXPO not working as it should, maybe worth a shot.

Yes, that's the other kit that I used to test this system. With that kit the system booted at 6000MHz without POST delays, but I just did 2 reboots after that and one cinebench run. It was stable during that time, but it may have been a coincidence, since this KIT I have also sometimes works well while from one moment to the next I can no longer POST, etc.

The only clear difference between kits was that my kit delays the POST (DRAM light) for >1 min with anything above 5200MHz, while that KIT that you mentioned booted without delays at 6000 MHz.

Unfortunately I don't have that other kit to do more extensive testing :'(

I will have to buy one and crossfingers... My fear atm is if this is caused by the CPU or mobo :'(

Download and run memtest

If you can send it back and exchange the memory for something like your friends.

Which one?

I used HCI memtest with 16 instances (16 threads), and after 300% there was no errors found (at 5200 MHz). Even so, this morning I went to turn on the machine and it gave me BSOD DPC WATCHDOG VIOLATION, so it seems to me that it's still unstable.

I have also ran PRIME95 Blend for 30 min, and OCCT CPU LINPACK (1hr) without errors at 5200 MHz. I will be doing more stress tests.
 
Last edited:
I will have to buy one and crossfingers... My fear atm is if this is caused by the CPU or mobo :'(
This is a subjective opinion obviously, but for me DDR5 is still really new and the issues haven't been 100% ironed out, so there's some compatibility issues between CPUs, boards and RAM. If your PC is otherwise stable, I wouldn't worry too much, especially as the X3D doesn't particularly care about RAM speed. It might be that a future BIOS fixes the issue.

The other kit is on the QVL and this one is not, so that could suggest something about the compatibility of these sticks with your setup.
 
Can you actually see the difference in running 5200 vs 6000? As Teras said x3d don't care much for ram speed, this would go along with my research over last few months before buying this chip, is strange as over the years everyone was saying amd prefer faster ram and then this.
 
Do you know what ICs those kits use?

From what I've read on another forum, the C36 kits are more likely to use Micron and some AM5 boards seem to be dissatisfied with this combination and struggle to train/optimise the board for them. Whereas, the C30 kits are more widely supported and the quoted kit (CMK32GX5M2B6000Z30) is present on the QVL of your motherboard.

I have Corsair CMH32GX5M2E6000Z36 Manufactured by Micron and with EXPO Enabled encounter Orange DRAM Light for 2 Minutes Each Boot!

It does actually boot though.

FYI :D
 
Can you actually see the difference in running 5200 vs 6000? As Teras said x3d don't care much for ram speed, this would go along with my research over last few months before buying this chip, is strange as over the years everyone was saying amd prefer faster ram and then this.

I understand, the difference is minimal in most of the cases.

The thing is that even at 5200 I'm not sure the system is stable. I had a BSOD today during boot at 5200MHz, fresh install so I highly doubt it was caused by any driver / system issue (maybe it was and was just coincidence).

Note: I have been running Prime95, OCCT linpack, OCCT memory test, HCI Memtest and so far, so good, even with the BSOD this morning... (5200).
 
Back
Top Bottom