Bug in BIOS with ECC RAM?

[Heiner-F]

I've noticed that the Z240 SFF reports an incorrect memory width when using ECC RAM.
However, when I move the same CPU and RAM into an HPE ProLiant ML30 Gen9 Server, it reports the correct values.
Both systems utilize the same Intel C236 Chipset.

The Z240 SFF is running the latest BIOS version.
The correct memory width should be 64 bits for data plus an additional 8 bits for ECC, totaling 72 bits.

When running 'dmidecode -t memory|grep -iE "ecc|width",' it should ideally return the following:

Error Correction Type: Multi-bit ECC
Total Width: 72 bits
Data Width: 64 bits

However, the Z240 SFF returns the following:

Error Correction Type: Single-bit ECC
Total Width: 128 bits
Data Width: 64 bits

I've conducted tests with the following CPUs: XEON E3-1240 V6, XEON E3-1270 V5, and i3-6100.
The RAM used is 4x DDR4 Samsung M391A2K43BB1-CTD (ECC UDIMM)."

EDIT: dmidecode (Linux) and wmic (Windows) report the same incorrect memory width!

[DGroves]

there is nothing wrong with the two systems differing in the memory width ,,............as each system has different memory channel configurations

the z240 only has two memory channels that are 2 dimms for each

the proliant G9 has 24 dimm slots with each channel comprising 4 dimm slots

this means each system's memory subsystem is completely different and as such you are trying to compare apples to oranges

[Heiner-F]

That is incorrect. I am talking about the tower server - HPE ProLiant ML30 Gen9 Server
It has exactly the same memory configuration as Z240 SFF. - 4 DIMMs (2 Channels/ 2 DIMMs per channel)

https://www.hpe.com/psnow/doc/c04834991

[Heiner-F]

I booted up the Z240 SFF with Windows 10 to check if there were any differences.
It's the same error as in Linux, reporting the wrong information.

Powershell commands:
wmic memorychip get dataWidth,totalwidth
DataWidth 64
TotalWidth 128

wmic memphysical get memoryerrorcorrection
5 (Single-bit ECC)