"ECC ERROR DETECTED "on every 32GB ECC Module tested on Coffee-lake Platform

[C.C]

While I use 32GB(8Gbit Chip) DDR4 ECC module on Coffee-lake Platforms by using MemTest86 V8.2, I encounter a problem.
After test 32GB DDR4 module always show "ECC Error Detected" while 16GB DDR4 modules work normally without “ECC Error Detected”.
I exchange dozens of modules ,different DIMM slots ,and different Mother Boards, all of them show "ECC Error Detected" after any test.(as attachment screenshot)
Have you tested 32G ECC Modules on Coffee-lake platforms before? Did you encounter the same problem?
Thank you.

Tested Platform Info:
CPU: Intel XEON-E2144G
Mother Board: GIGABYTE C246-WU4 , ASUS C246-PRO
Program: Memtest86 V8.2

[keith]

That looks a bit strange. Can you send or attach a copy of the MemTest86.log located under EFI\BOOT\ of the USB drive.

[C.C]

That looks a bit strange. Can you send or attach a copy of the MemTest86.log located under EFI\BOOT\ of the USB drive.

Hello Keith,
The attachment is the MemTest86.log file

Thank you.

[keith]

Thanks for the logs.

There doesn't seem to any issue with the logic for detecting ECC errors, so I'm suspecting a hardware bug with the CPU.

2020-07-31 06:45:01 - ERRSTS=0003
2020-07-31 06:45:01 - ERRLOG0[0]=008D0003
2020-07-31 06:45:01 - ERRLOG1[0]=0001FC00
2020-07-31 06:45:01 - [MEM ERROR - ECC] Test: 4, (Col,Row,Rank,Bank): (0,1FC00,0,0), ECC Corrected: yes, Syndrome: 008D, Channel/Slot: 0/0

The log entries indicate there was both a correctable single-bit and uncorrectable multi-bit ECC error. Uncorrectable ECC errors usually result in a system halt which doesn't appear to be the case.

Is the same CPU being used for each motherboard for the tests? Or different ones?

[C.C]

Hello Keith,

Yes, the test is using same CPU.

[David (PassMark)]

This seems to be a pre-release CPU, which we don't have access to.
Genuine Intel(R) CPU 0000 @ 3.00GHz
So reporting a fault to Intel will be hard for us. Do you have a contact in Intel you can talk to?