Stuck DRAM LED on warm reboot

[Pete2]

I recently built my first AM5 system, upgrading MB, CPU and RAM. All new parts:
- Asus TUF Gaming B650-PLUS

- Kingston DDR5 6000 CL30 KF560C30BBEK2-32 (2x16 GB kit) EXPO certified. On MB's QVL.

- 7800X3D

 

Windows 11 latest build and updates. Latest AMD Adrenalin & Chipset drivers. Latest BIOS as of now: v3222.

 

Existing parts: Corsair Rm750x PSU, 2 x M.2 SSD (gen4). GPU Powercolor RX6800XT Red Devil, Cooler Noctua NH-D15s with additional fan. All these parts were working fine with my previous B550-PLUS and 5700X.

 

Initial setup went fine, no crashes or bsods, bootup failures. I even gamed on it a little bit without issues. I also ran some benchmarks like 3DMark CPU Profile, TimeSpy Extreme, Cinebench r23 & 2004. All went fine, no crashes. I read that enabling Memory Context Restore would improve POST times. So I did that + set the Power Down enabled as well. At this point EXPO 6000 was still enabled.

 

One day I ran Cinebench r23, just one multicore run and shortly after that I rebooted and then it got stuck on DRAM LED. Hit reset button, no still did not POST. After a few times it posted, went to BIOS immediately and one of the memory sticks was missing. Swapped out the sticks, tried 1 stick in both slots etc. all combinations. Some combinations would randomly boot and I also reset BIOS at some point (at this point EXPO was disabled). Initial boot would succeed but then reboot from windows did not.

 

I ran several memory tests without errors: Windows Memory Diagnostic, Memtest 86, TestMem5. Sometimes it would POST with 1 stick even though 2 was installed and so 16 GB memory showed up but TestMem5 somehow detected 32 GB of memory but couldn't allocate over 16 GB of it.

 

So then I switched Memory Context Restore back to Auto. Still no help. Finally I thought to turn the Memory Context Restore to Disabled. Leaving the Power Down to Auto. And now it reboots every time. 

 

The questions here is, is something faulty? I mean, Memory Context Restore should be safe option at Auto, right? And also running the RAM on JEDEC and since the Memory Context Restore setting is on Auto by default and it still doesn't boot with that... What is Auto anyway?

 

I found this old thread, in which the OP had a faulty CPU, almost the same issue: 

 

 

I would of course switch parts but I don't have any and would have to spend money to get test parts. So I'm asking for comments, what do you think is faulty here?I'm hoping I could narrow down the issue without buying a bunch of new parts. Thanks in advance.

 

 

[Bjoolz]

34 minutes ago, Pete2 said:

 

The questions here is, is something faulty? I mean, Memory Context Restore should be safe option at Auto, right? And also running the RAM on JEDEC and since the Memory Context Restore setting is on Auto by default and it still doesn't boot with that... What is Auto anyway?

Memory context restore is that it loads the previous settings it found stable during memory training. On is always load, Auto is load unless there is an issue (Though some boards do it differently where Auto can mean that it doesn't load the settings if XMP is enabed). With Auto, it should reset the settings if it was deemed not stable during POST. 

[Tetras]

1 hour ago, Pete2 said:

The questions here is, is something faulty? I mean, Memory Context Restore should be safe option at Auto, right? And also running the RAM on JEDEC and since the Memory Context Restore setting is on Auto by default and it still doesn't boot with that... What is Auto anyway?

It doesn't seem to work correctly 100% of the time. I've seen a fair few posts where the OP said "had to turn it off, doesn't work/system unstable". If your PC is fine with it off, I wouldn't worry about it.

[Pete2]

16 hours ago, Bjoolz said:

Memory context restore is that it loads the previous settings it found stable during memory training. On is always load, Auto is load unless there is an issue (Though some boards do it differently where Auto can mean that it doesn't load the settings if XMP is enabed). With Auto, it should reset the settings if it was deemed not stable during POST. 

That makes sense but in this case if it doesn't POST when on Auto it won't reset settings either. And I have waited long enough, over 4 minutes. So it seems it (BIOS?) doesn't detect the issue and after a forced power off and reboot, it boots fine again.

[Bjoolz]

On 4/29/2025 at 6:13 AM, Pete2 said:

That makes sense but in this case if it doesn't POST when on Auto it won't reset settings either. And I have waited long enough, over 4 minutes. So it seems it (BIOS?) doesn't detect the issue and after a forced power off and reboot, it boots fine again.

You could try updating the BIOS. 

[Pete2]

Thanks all, I think it's good now. I have disabled MCR on both locations on BIOS and when it posts it's stable (ran TM5 overnight with no errors). On some random warm reboots it MAY fail memory training and go to failsafe mode according to BIOS. Running EXPO II mode now since I read it may be more reliable since it defines more values as fixed rather than Auto. 

On ZenTimings one value, tRDWR differs between sticks: one is 21 and the other 22.