Help! ProArt X570 Creator - 5950X - DRAM Yellow light

[DeadlyMixStudio]

Swapped out my MSi MPG X570 Gaming Plus for an ASUS ProArt X570 Creator WiFi.
The problem is that after a minute or 2 the DRAM debugger light turns on, then it gets extremely slow until it finally crashes. Sometimes the displays go black and other times I get the light blue "Oops Windows took a dump... 0%" screen.

I'm gonna try taking out the single 32GB 3600MT/s CL18 DDR4 stick of Kingston FURY Renegade RGB and swap it with one of the basic 16GB 2666 sticks I have on-hand and see if that makes a difference tomorrow.

Im just hoping someone has more insight into this because I need to get the recording studio up and running asap.

Specs:
• Windows 10 Pro x64
• ASUS ProArt X570 Creator WiFi
• Ryzen 9 5950X
• Fury Renegade RGB 32GB 3600MT/s CL18
[1 stick now, but I have 3 more to install once it's running]
• EVGA GT 710 gpu
• Focusrite Pro RedNet PCIeR Dante AVoIP interface card
• MSi MPG Coreliquid K360 cooler
• ASUS RoG Thor 1200W w/OLED
• 1TB M.2 NVMe (top slot)
• 1TB SATA SSD
• (3) 1TB USB3 SSD (behind back panel in case)
• Thermaltake View 37 ARGB (case)

I know some of those specs are unnecessary but you never know what small detail might lead to a solution.

[xeromist]

What speed were you running the 3600 stick? You can run the same stick at a lower frequency to see if that makes a difference. Once you have an apparently stable speed you should still run some memory tests to make sure you don't get surprise crashes in the middle of your work later.

[DeadlyMixStudio]

xeromist wrote:
What speed were you running the 3600 stick? You can run the same stick at a lower frequency to see if that makes a difference. Once you have an apparently stable speed you should still run some memory tests to make sure you don't get surprise crashes in the middle of your work later.

I was letting it stay at the lower frequency it defaulted to. Either 2666 or 3200, I don't remember specifically.

Today I swapped it out for a stick of TeamGroup 32GB 3200 CL20 (no rgb) dual-channel DDR4 I had on-hand.
- Same result

Then I tried a 16gb Crucial 2666 CL16 (no rgb, no XMP) in the 2nd, 3rd, and 1st slot from the CPU.
- Same result, but seemed to take longer before quickly but gradually slowing to a crawl and finally crashing

** All of today's troubleshooting done in BIOS only. Did not attempt to enter Windows.

I was able to poke around in BIOS long enough this last time to find what may be the cause.

BIOS > Advanced Mode > Monitoring > Chipset...
Bright red letters: "Chipset 104°c" (215°f)

I've been able to get our IT guy on the phone off & on throughout the day today and he told me to reinstall the CPU, look everything over, and check it again using the 16gb 2666 Crucial stick only in the 2nd slot from the CPU again.

Overheating makes sense because it would also explain why I have less and less time to look through BIOS each time unless I turn it off for awhile between attempts.

I didn't expect overheating to be an issue with this mobo. I assumed that, because it's passively cooled, they had come up with a design that better prevented it from happening.

Everything is still set to default. I haven't even had the chance to install drivers and check for updates yet, so configuring performance isn't yet on my to-do list

[xeromist]

As a test you could try propping up a fan to blow on the chipset. If that's the problem then you should notice a difference (unless the heatsink is making no contact at all).

[DeadlyMixStudio]

I flashed the BIOS update from April 2022 (mobo manufactured March 2022).
I depopulated the mobo to flash it by itself, because I was worried it might overheat and crash during the update otherwise.
Both the button and way the light works on the ProArt X570 are a little different from most mobos I've seen so, unable to see what's going on, I panicked. The lights blinked 3 times after I released the button then stayed solid. I waited awhile for it to turn off then grabbed the tower, stuck a GPU + CPU + RAM in my pocket and drove it straight to our IT guy.

He put it back together and looked at BIOS to see if it was working and if the update worked. It did.

Then he went into BIOS > Advanced > Monitor > Temperature > Chipset.
For the first time, it wasn't red. However it was still around 200°F/above 90°C.
He pushed his hand to the heatsink/cover to see if it was really that hot and it didn't feel hot.

His theory was that the sensor is bad, so it just thinks it's getting hot and shuts down. to save itself from the heat that's not actually there.

That seems possible but I gave it some thought and I figure another possibility is that it IS that hot but the reason is that the heatsink isn't tight enough to make enough contact to conduct the heat away from the chipset. That, I believe, would end in the same result. It would read that it's hot because it is hot but it wouldn't feel hot because the heatsink isn't drawing heat from the chipset.

To test whether the issue is a bad sensor or loose heatsink, I aimed a large room-fan at the open case.

Unfortunately, I can't be certain of the results.
The temperature was at 89°c and climbing when I decoded to point the fan at it.
It reached 95°, pointed the fan and it stopped rising within 2 seconds and stayed there.
Several seconds later, the temperature dropped every 10 seconds or so until it reach 92°c and stayed there.

I waited a couple minutes. It didn't change.
I decided to pull the fan away...
The temperature went back up and hit 98°c before I aimed the fan at it again. Then it did basically the same thing.
Stopped rising but stayed at 98°c for about 30 seconds.
Then slowly dropped to 94°c and stayed that way for a couple minutes before I decided to try exiting BIOS to run Windows 10 Pro x64.

I kept the fan aimed at the mobo.
It let me know I forgot to plug the CPU fan back in, so I plugged it in, it rebooted, then took over 5 minutes to boot Windows before crashing..

My theory is that the fan wasn't enough to keep it below 100°c for any operation beyond running BIOS without making changes.

I THINK that the heatsink isn't tight enough and it actually is getting hot as a result BUT my test was too brief and the results not identical enough both times (staying at 1 specific temperature when the fan was aimed would've been more twlling) to be certain that the fan was the difference beyond any doubt.

I REALLY don't want to keep my business shutdown for 6 weeks or however long waiting for ASUS to repair and return the mobo. So I'm hoping to find a nondestructive way to get to the heatsink and tighten it to the chipset better myself.

[xeromist]

I gather this MB is not a recent purchase then? The most expedient resolution is usually an exchange with the retailer.

[knighdeath]

Try you 3800 memory.

Change to D.O.C.P.

Change FCLK Frequency : AUTO to 1800 (1800 x 2 = 3800).