Question Supermicro X9DRD-IF

[Karliczyniak]

Hi all,

I have a very strange issue, or potentially I am doing something very wrong.
Recently I purchased the following hardware:

2x Intel Xeon 2650v2's
1x SuperMicro X9DRD-IF mainboard
128GB DDR3 ECC


The problem is that 2 of the DIMM slots don't work. In total, it has 8 Dimm slots (4 for each CPU).
I can populate 6 of them (96 GB of ram) but as soon as I put ram into the other slots, I get a B7 post error.

If I remove all Ram and only have one stick in one of these slots, the same thing B7 post error. So, the Dimm slots are faulty, right? well, I don't think so. I contacted the seller of the board who replaced it with another for me. I have the EXACT same issue.

I can take a working 16GB Ram, put it into any DIMM slot (except these 2 that aren't working) and it works. Take the exact same 16GB of ram and put it into one that isn't working and it posts with B7 error. I have swapped the CPU's with 2 others (again 2x 2650V2's and also 2x 2650L V2's and the same thing). I've even tried with one CPU, no luck.

I've updated to the latest firmware. Changed the CMOS battery, reset the CMOS, same thing.
I've tried the board in the case and also outside of the case (in case of a grounding issue), same problem.

I have no idea what to do next.

In summary, I've tried 8 different CPU's, 20 different sticks of ram and 2 mainboards with the same issue. I know there must be some trick to getting all 128GB of ram to work but I just don't know...

 

[COLGeek]

Which slots are the ones in question?

https://www.supermicro.com/manuals/motherboard/C606_602/MNL-1488.pdf

Is all of the memory 100% matching?

 

[Karliczyniak]

Hi there ColGeek. Thank you for the reply.

Yes, all the memory is matching.

The 2 slots are the 2 on the farthest left on page 1-2 of the link you post (C1 and D1).

 

[COLGeek]

I came across several posts on other sites where others "fixed" the problem by re-seating the CPU.

As you have swapped multiple CPUs, this would seem like a moot point....except, have you tried swapping just the CPUs with all DIMMs populated to see if the error "moves" to other slots?

Also, can you access the BIOS and see how it shows the installed memory?

 

[Karliczyniak]

Hi again,

The error doesn't move to other slots. I have had all populated, swapped the CPU's and same b7 post error.
I've also read the same threads, re-seating the CPU's does nothing to help the situation... unfortunately.

When I access the bios, it shows the CPU's without any issues and also shows the ram slots populated which are populated (all except the ones it doesn't work in ofcourse). It shows the 2 slots "empty" when they are empty which maybe means that it knows the DIMM slots exist?

 

[COLGeek]

If memory is physically installed in those slots, does the BIOS see as empty?

 

[Karliczyniak]

Hi there.

I do not know, the motherboard gets stuck at POST error B7 when RAM is installed in either of those slots. Accessing the BIOS is not possible when RAM is in those slots.

At first, like I mentioned, I believed it to be a board fault, but upon getting a replacement board, the same issue exists which leads me to believe its my fault somehow.

 

[COLGeek]

Hi there.

I do not know, the motherboard gets stuck at POST error B7 when RAM is installed in either of those slots. Accessing the BIOS is not possible when RAM is in those slots.

At first, like I mentioned, I believed it to be a board fault, but upon getting a replacement board, the same issue exists which leads me to believe its my fault somehow.

This is definitely an odd situation. What common devices have been used with all the variations you have tried? I assume storage devices and power supply. Anything else?

 

[COLGeek]

Also, are you running all of your memory settings (in the BIOS) in their default state? I am interested in what settings you have for Channel Interleaving and Rank Interleaving.

 

[Karliczyniak]

All in bios is default. Memory, although it is 10600, shows as 800mhz when auto detected.
Additionally, it runs in "independant" mode.

I'm going to try and make a video.

 

[COLGeek]

So both of the settings I asked about are set to AUTO?

 

[Karliczyniak]

yes

 

[COLGeek]

Are both CPUs and all memory 100% the same revisions? No variations or minor differences?

To verify, any memory configuration that includes memory installed in C1 or D1 trigger the B7 error. Correct?

 

[Karliczyniak]

Hi yes,

so I just checked. Both Rank and channel interleaving are "AUTO".

I've just installed 32gb ram in dimm A1 and B1 to get the info you want:

DRAM RAPL mode is mode1
Memory mode is independant
Current memory speed shows 800mhz

Any memory, even one stick by itself in either DIMM slot C1 or D1 trigger a B7 hang.

 

[COLGeek]

What is the make/model of the memory you are using?

Have you reached out to SuperMicro on this?

 

[Karliczyniak]

I am using Cisco branded ram with Samsung modules taken from a Cisco blade server.
I haven't reached out to Supermicro, no.

Like these: https://www.gekko-computer.de/en/Parts/Memory/DDR3-RAM-23986.html

 

[Karliczyniak]

Hmmm, I just read that this ram isn't suitable for PC's.....

https://www.gekko-computer.de/en/Parts/Memory/DDR3-RAM-23986.html

"Speicher ist nicht für PCs geeignet."

 

[Karliczyniak]

Is server blade ram different somehow? Or does it mean by PC's "not suitable for any normal PC and xeon server mainboards are excluded"... that is weird. I don't really know how to read that,

 

[COLGeek]

That seems to be memory made for Cisco devices. While I haven't looked, did you see if those were listed on the motherboard's QVL?

ECC memory is used in servers/workstations. Not typical consumer PCs. That is what that statement is telling you.

 

[Karliczyniak]

Unfortunately on their website, it doesn't state specifically.
However, if it posts using all the other DIMM slots, not sure why it wouldn't just in these 2 dimms.

 

[COLGeek]

Unfortunately on their website, it doesn't state specifically.
However, if it posts using all the other DIMM slots, not sure why it wouldn't just in these 2 dimms.

I don't either, but that would be the first thing SuperMicro would point out as a possible issue.

Do you have any other memory (different make/model) to test with?

 

[Karliczyniak]

Unfortunately, no 🙁

 

[COLGeek]

Unfortunately, no 🙁

Understood. Either run with only 6 slots or find some different memory to test with. Seems there are no other options at this point.

 

[Karliczyniak]

Thank you very much for all your assistance COLGeek, I really appreciate your time and efforts here. I will try and obtain some other ram to test with.

 

[COLGeek]

My pleasure. Please follow up when you can.