Unique (about to die) 9800X3D sample - physical and highly localized damage to DQ lines
(EDIT 1)
I have just sent GamerNexus an email. When I get reply I will attach it in my edit 2
(END EDIT 1)
Long story short. I was validating my theory that Proc Dq Drive strength and RTT WR reflections are causing some type of damage at the IOD die / contact pads due to continues stress which could potentially prove why some Zen 4 and Zen 5 are dying
To do so I created a custom memory stress test which basically spams write operations from CPU to RAM. The voltages were normal, nothing extreme.
This sample is unique in a sense that the chip itself is still perfectly bootable, which could be of great value to anyone researching the topic of 9800 X3D's dying
-
After 16 hours of stress testing the system shut itself down → red light → cold boot → 4 hours stress test → red light → cold boot → 2 hours stress test → red light
Afterwards I took the CPU out of the socket to inspect the pins. I found highly localized damage at MBB_Data 20 and MBA_Data 12. Where pin MBB DATA 20 is even missing the majority of its substrate
These pins are governing bidirectional read and write operations, which proves my theory behind continuous stress during write operations causing some type of degradation
-
Here is a list of exact pins which shows some type of damage to the substrate of the pads.
Left is position in accordance to the AM5 pin layout found here. Right is the name of the affected pin https://commons.wikimedia.org/wiki/File:Socket\_AM5\_pinmap.svg
(Left bottom part of the chip)
F31 → MBA_Data 12
L31 → VSS
E27 → VSS
(Right bottom part of the chip)
This area is the bronze/dark circle seen of the photo. Pin with dark spot (MBB DATA 20) on it is missing the majority of the substrate)
AW29 → MBB_Data 20
AP29 → Test 47_2
AT29 → VSS
AP31 → Test_CCD0
AW27 → VSS
-
My DM's are open for anyone which has the means and the knowledge to analyze the damage to the substrate and could somehow tie it (or not) to 9800 X3D failures