Memory slot failed !?

[hlefranc]

Hello,

I have a HP microserver Gen8 with CPU G1610T with 8Go + 4Go that worked fine. 

I upgraded successfully the CPU to E3-1265 v2, after that I can start the the server only on one memory bank with the 4Go or 8Go bank.

I have the question, normally we should be able to start the server using the slot 1 or 2 using one memory bank, right ?

Just to know if the memory slot  is out of service or not ....

Regards,

[Rakshitha_K]

Hi @hlefranc 

Yes, I believe that Gen8 server should boot with a single RAM module in either DIMM slot.
If it only boots when a module is in one particular slot after your CPU swap, that points to a compatibility/BIOS issue or a hardware fault introduced during the upgrade, not an intended restriction of the board.

What normally happens is that the Gen8 motherboard has two DIMM sockets (one per memory channel). It will boot with a single compatible UDIMM in either socket, though there is a “primary” socket the firmware prefers on some boards. You do not need both modules present for POST.

Why this behaviour can happen after a CPU change:
   - Different CPUs can change memory controller behaviour (timings, supported DIMM types) and some CPUs require a BIOS update to work properly with certain memory.
   - The act of swapping the CPU can also disturb the socket, pins, or nearby components (bent pins on LGA, mis-seated CPU, cooler pressure) and cause one DIMM slot not to be seen.
   - Memory compatibility: ECC vs non‑ECC, rank, or speed differences may cause odd behaviour with a different CPU/fused microcode.
   - Firmware/BIOS: older BIOS might not include microcode or support for the new CPU leading to odd memory detection.

Suggested troubleshooting steps:
   1. Re-seat CPU and cooler carefully and recheck for bent pins on the CPU socket (LGA style on Gen8). Reseat the heatsink.
   2. Update the server BIOS/UEFI to the latest Gen8 BIOS (from HP) while using a CPU that allows the update if necessary.
   3. Clear CMOS/NVRAM: power down, remove power, remove the coin cell for a minute (or use the clear jumper) then reboot.
   4. Test memory modules individually:
       - Test DIMM A alone in slot 1.
       - Test the same DIMM alone in slot 2.
       - Repeat with the other DIMM.
        If a known-good DIMM works in slot X but not slot Y, that points to a bad slot or socket mechanical issue.
   5. Try the original CPU (G1610T) and confirm whether both slots work again. If both slots work with the old CPU, the new CPU or its seating/compatibility is likely the cause.
   6. Check HP iLO/LED POST codes or beep codes for memory errors (the front panel/system health LEDs can help identify memory errors on Gen8).
   7. Verify memory specifications: make sure the DIMMs are supported types (unbuffered UDIMM, supported speeds, ECC vs non‑ECC as your motherboard supports).
   8. If you determine a fault with a single slot and it’s not caused by seating or BIOS, the motherboard socket/traces may be damaged, plaese consider repair or RMA.

Regards,
Rakshitha_K

[hlefranc]

Hello @Rakshitha_K,

Thank your for this feedback and these useful advices.

I'm going to remove all the compoments from the motherboard, check all the slots and CPU socket pins.

Indeed, perform a rollback to the old processor to check if the origial configuration still works fine is a good idea.

About the BIOS, what should be the right version for this type of processor ?

Regards,

[hlefranc · Accepted solution]

Hello  @Rakshitha_K,

Good new!

After to replaced the CPU Xeon 1265L v2 by the orginal CPU, I can boot to the memory slot 1 or 2 with the bank 8Go  or/and 4Go.

After that, I changed again the original CPU by the Xeon 1265L v2 by being careful and now that works fine with 2 memory banks!

I think I overtightened the radiator screws or locked the CPU lever too tightly ...

Many thanks for your help.

Regards,

[Thaufique_Mod]

@hlefranc,

Glad to know your concern has been addressed! 
Happy to help in any way we can.