ALTRAD8UD-1L2T - Micron 128GB RAM stick issue

Just finished my ALTRAD8UD-1L2T build! I was really excited but my system doesn’t boot. :frowning:

Looking at the BMC logs, it seems like it doesn’t like the RAM module:

NOTICE:  DRAM FW version 220414
CP: 0ff00900
MEMC param:
  mcu_enable_mask = 00000001 [default: 000000ff]
MCU param:
  ecc_en = 00000002 [default: 00000001]
DRAM populated DIMMs:
  SK0 MC0 S0: LRDIMM[2c:80] 128GB 3200 ECC 4R x4 RCDDB[32:86] 72ASS16G72LZ-3G2F1  
CP: 3ff00100
CP: 3ff00200
CP: 00f01902
CP: 0ff00a00
CP: 0ff00b01
CP: 00f01901
CP: 00f01904
CP: 00f02901
CP: 00f01905
CP: 00f02902
CP: 00f02903
CP: 00f02904
CP: 00f01907
CP: 00f01906
CP: 00f0190a
CP: 00f01908
CP: 00f01909
CP: 0ff00b02
CP: 0ff00c00
CP: 3ff00300
CP: 00f01b00
CP: 00002b01
CP: 00002b03
CP: 00002b05
CP: 00002b02
CP: 00002b04
CP: 00102b01
CP: 00102b03
CP: 00102b05
CP: 00102b02
CP: 00102b04
CP: 00202b01
CP: 00202b03
CP: 00202b05
CP: 00202b05
CP: 00202b05
CP: 00202b05
ERR: MC0:R2: MREP Train (Retry) FAIL
ERR: 80f0f190
CP: 00302b01
CP: 00302b03
CP: 00302b05
CP: 00302b02
CP: 00302b04
ERR: 80f0f160
ERR: 8ff0c700
CP: 3ff00800
ERR: bff0c100
ERROR:   SK[0]: 00120001
ERROR:     MC[0]: 00f100b4
ERROR:   DDR initialization failure. Enter safemode...

My understanding is that this ram stick should work, since it is a LRDIMM module.

Please let me know if there’s any chance I can make it work with this RAM module.

Thanks!

You hate your system so you used just one memory stick?

64 cores, 8 memory channels, 1 memory stick? I see pattern.

And if it’s on the memory QVL.

My system uses a non-qualified RDIMM memory module, which works well. However, according to one of my friends who used to work in Ampere, you should use memory from the Qualified Vendor List (QVL)

2 Likes

Yeah, some sticks which aren’t on the QVL might work. From my time getting my system set up, it’s kind of a game of luck.

Yeah, ime it’s a much smoother time getting the system to POST when using the QVL.

Not being in the US, it is quite hard for me to find any of the RAM on the QVL (at a reasonable price).

The 128GB stick I got for a big discount, but I’m still happy that I didn’t order more, since it doesn’t seem to be working on this system.

How about running a parallel, unofficial thread with the non-qualified RDIMM that people have tried and it worked?

@quocbao could you please share that part number?

1 Like

This is mine

root@voyager:~# lshw -c memory | grep bank -A2
     *-bank:0
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43DB3-CWE
--
     *-bank:1
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43DB3-CWE
--
     *-bank:2
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43EB3-CWE
--
     *-bank:3
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43DB3-CWE
--
     *-bank:4
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43DB3-CWE
--
     *-bank:5
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43EB3-CWE
--
     *-bank:6
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43DB3-CWE
--
     *-bank:7
          description: DIMM DDR4 3200 MHz (0.3 ns)
          product: M393A2K43DB3-CWE

Cool, thanks! Yours at least is on the Ampere Altra AVL - March 2024.