Need help my Ryzen 5 computer doesn't want to post at random

[AdamGrant82]

I'm having a random problem with the computer not posting with the CPU debug LED lit after a forced restart it posts but now I'm getting a message that says "Memory over clock has failed" but I only turned on XMP I have the memory voltage set to 1.35V I had to manually set the timings of the memory kit to match what they are rated for but it seems like 3200MHz and even 3133MHz is a no go it's stable with memtest86 at 3200 but still won't post at random with the CPU debug LED on and there's only one short beep once the CPU debug LEDs stays on but other times there no beeps at all and no picture on the screen I'm lost it can't be the memory if it passes memtest86 right?

I ran a full system stress test for over two hours and it always passes even with the memory at 3200MHz I never had a computer fight me this much before. I could put the old memory kit back in but even that kit won't run stable at rated speed no matter what motherboard is used. I have already replaced all the hardware in my computer besides the CPU and I was having random memory related problems with this CPU in my other system. 

Once the computer's booted into Windows it works fine it's only that random time once the computers been off all night could I have a CPU problem? The memory kit is supported by my motherboard I checked on the MSI's site for my board.

Spec's:
AMD Ryzen 5 3600 
G.SKILL TridentZ RGB 32GB 3200MHz Memory
MSI Radeon RX 5600 XT GAMING MX 6GB
MSI MAG b550 Tomahawk Motherboard
MSI MAG CORELIQUID 240R AIO Liquid Cooler
Seasonic FOCUS GM-750, 750W 80+ Gold PSU
Samsung 980 Pro 500GB PCIe Gen 4 SSD
Western Digital Blue 4TB hard drive(upgrading to 8TB)
Creative Sound BlasterX AE-5 Plus Sound Card
Windows 10 Pro
Case:
Phanteks Eclipse P400A Digital RGB Black
Cooling Fans:
Phanteks SK120 DRGB PWM FAN X4
 

[Analog]

Seems there are is an issue with the motherboard not being able to train the memory at those speeds. If this is indeed the case you are left with two options. First option is to run without XMP, but instead with the standard JEDEC speed which I assume is 2400MHz. The second option is to try and fiddle with the memory timings, possibly loosen then up a bit more and hope that this will make the system stable. Out of curiosity, is this G.Skill kit in the QVL for the board that you have? 

[AdamGrant82]

40 minutes ago, Analog said:

Seems there are is an issue with the motherboard not being able to train the memory at those speeds. If this is indeed the case you are left with two options. First option is to run without XMP, but instead with the standard JEDEC speed which I assume is 2400MHz. The second option is to try and fiddle with the memory timings, possibly loosen then up a bit more and hope that this will make the system stable. Out of curiosity, is this G.Skill kit in the QVL for the board that you have? 

I had memory related problems with two different motherboard and two different memory kits the only part that's the same is the CPU and yes the memory kit is listed as supported I made sure before I ordered it but I found another post this guy was having the same problem right down to the CPU debug LED and not posting and the comments are saying it's the CPU is on it's last legs and could die soon I had the CPU Vcore set to auto for most of the time I had it and auto seems to set the CPU core voltage to 1.4-1.43V I changed it to offset mode with auto voltage and a -0.100 offset after it didn't post this morning I went back into the settings and changed the offset to -0.200 and lowered the memory to 3000MHz running Prime95 now and over a hour and it's fine the temps went way down on auto settings the temps were at 70C-76C under full load now while running Prime95 it's at 50C-53C 

Here's the link to the other post

https://pcpartpicker.com/forums/topic/346646-pc-wont-boot-cpu-ez-debug-led

[Analog]

1 hour ago, AdamGrant82 said:

I had memory related problems with two different motherboard and two different memory kits the only part that's the same is the CPU and yes the memory kit is listed as supported I made sure before I ordered it but I found another post this guy was having the same problem right down to the CPU debug LED and not posting and the comments are saying it's the CPU is on it's last legs and could die soon I had the CPU Vcore set to auto for most of the time I had it and auto seems to set the CPU core voltage to 1.4-1.43V I changed it to offset mode with auto voltage and a -0.100 offset after it didn't post this morning I went back into the settings and changed the offset to -0.200 and lowered the memory to 3000MHz running Prime95 now and over a hour and it's fine the temps went way down on auto settings the temps were at 70C-76C under full load now while running Prime95 it's at 50C-53C 

Here's the link to the other post

https://pcpartpicker.com/forums/topic/346646-pc-wont-boot-cpu-ez-debug-led

On auto it is normal for the processor to idle at around 1.4V. I have even seen some 3950X chips idling at around 1.46V. This shouldn't be a problem though as this voltage is not applied under load. Due to the LLC the voltage will droop significantly when under load to a bout 1.2-1.25V, which again is completely fine. This being said manually setting the VCore to 1.4-1.43V for an all-core overclock will signifantly degrade the processor to the point where it will be unstable. Just to rule out memory you can run HCI Design memtest and let it run to a minimum of 400%, even better let it run overnight (10-12h) and see if it finds any errors. Keep in mind this little program is x86 based and thus cannot test more than 3.5GB in one instance. My advice would be to open as many tabs of it as threads that you have and put in an equal amount of RAM to be tested in each. Just fyi, keep about 1.5GB reserved for system tasks. Meaning you have 32GB, I would test 30.5GB or better yet 30GB. All of this is in the instructions for the program but just saying, because it is important. Link for the application here: https://hcidesign.com/memtest/ .

[LogicalDrm]

-> Moved to Troubleshooting