Fujitsu server memory modules – registered ECC, but still won’t POST

configurationeccmemory

I have a Fujitsu TX150 S7, for which I want to upgrade the memory. I thought that would be simple enough, but apparently that is not so.

The manual for that server states the following:

Memory slots: 6
Memory slot type: DIMM (DDR3)
Memory capacity (min. - max.): 1 GB - 32 GB
Memory protection: ECC

along with some notes:

Memory notes: For the following modules a maximum of 4 DIMMs can be configured: 
4GB (quad-ranked) with 1066MHz, 8GB RDIMM, all UDIMM modules.
Maximum configurable memory capacity: UDIMMs (max. 16GB) with all processors; 
RDIMMs (max. 32 GB) with Xeon CPUs only.
The following modules can be configured 6x: 4GB (dual-rank) and 2GB (dual-rank)

The manual then goes on telling about specific configurations that are supported:

Memory modules:
     1 GB (1 module(s) 1 GB) DDR3, unbuffered, ECC, 1333 MHz, PC3-10600, DIMM
     2 GB (1 module(s) 2 GB) DDR3, registered, ECC, 1333 MHz, PC3-10600, DIMM
     2 GB (1 module(s) 2 GB) DDR3, unbuffered, ECC, 1333 MHz, PC3-10600, DIMM
     4 GB (1 module(s) 4 GB) DDR3, registered, ECC, 1066 MHz, PC3-8500, DIMM
     4 GB (1 module(s) 4 GB) DDR3, registered, ECC, 1333 MHz, PC3-10600, DIMM
     4 GB (1 module(s) 4 GB) DDR3, unbuffered, ECC, 1333 MHz, PC3-10600, DIMM
     8 GB (1 module(s) 8 GB) DDR3, registered, ECC, 1066 MHz, PC3-8500, DIMM

From that I gather that I need to get ECC memory, but since I have a Xeon CPU (Intel(R) Xeon(R) CPU X3450), I can get either type of ECC memory: Registered or Unbuffered, as long as I don't mix them.

As a result, I picked the following memory (times six):

Samsung 4GB 2Rx4 PC3 - 10600R - 09 - 10 - E1 - P0

From that I can read that it's a dual-ranked 4 GB memory bar, it's registered, I'd assume it's ECC because it was sold as ECC memory, and because the RX300 S6 that needs ECC memory would work with it, and its speed is PC3-10600, which matches what's in the manual.

However, the moment I put even one of these memory bars (or two, or four, or all six) into the server, it will no longer POST. It'll hang, with nothing on the screen, and write the following error in its event log after a while:

BIOS POST Watchdog - Action: Hard Reset (Post Code: 0xC0)

The same memory would work perfectly fine on another server (a Fujitsu RX300 S6). And likewise, if I put the 2 GB unbuffered DIMMs that the TX150 S7 came with back in, it'll work without any problems.

I can't find out why that's the case – where is my understanding of server memory compatibility wrong?

Best Answer

After trying various types of memory, I am convinced that the memory controller of the Fujitsu Primergy TX150 S7 chokes on the databus-width of the individual DRAM chips.

The memory chips that the server came with, and that ultimately worked, all had a databus width of x8 (2Rx8).

That you can't mix ranks (1R and 2R), ECC, buffered and unregistered memory was clear, but I did not think that the databus width of the individual chips would matter. Up to now I thought that less databus width simply meant more chips on the RAM bar (and most sites I found on the internet describe the difference between 2Rx4 and 2Rx8 as being simply more/less chips on the RAM, and no other effect).

However, none of the memory chips with x4 (2Rx4) work in the TX150 S7, even though they work fine in other servers, and any of the x8 (2Rx8) do work, making it fairly obvious.

Related Topic