I'm completely stuck and would really appreciate input from people familiar with SP5, ASRock Rack, EPYC Genoa, AST2600/BMC behavior, or server power sequencing.
Hardware tested
Motherboard #1
ASRock Rack GENOAD8X-2T/BCM
Motherboard #2
ASRock Rack TURIND8-2L2T
CPU #1
AMD EPYC 9254 (Genoa, 24-core)
CPU #2
AMD EPYC 9254 (different physical CPU, different seller)
RAM
SK Hynix DDR5 ECC RDIMM
64GB
HMCG94MEBRA103N
PC5-4800B
PSU #1
Seasonic PRIME TX-1600 Noctua Edition
PSU #2
Another TX-1600 unit with different cables
Symptom
The symptom is IDENTICAL across both motherboards and both CPUs.
BMC works perfectly:
IPMI Power On gives:
Performing Power Action.. Please Wait
Retrying...please wait. Retries Left : 2
Retrying...please wait. Retries Left : 1
Performing power action failed.
Host never enters ON state.
What I tested
Motherboards
CPUs
Power supplies
Memory
Build configuration
Cooling
CPU installation
TURIND8 sensor page
Only standby rails exist:
System Inventory
Processor Info:
Information Not Available
No CPU information shown.
LEDs
Why I'm confused
At this point I have changed:
That makes me think this is either:
Has anyone seen:
Any ideas appreciated.
Hardware tested
Motherboard #1
ASRock Rack GENOAD8X-2T/BCM
Motherboard #2
ASRock Rack TURIND8-2L2T
CPU #1
AMD EPYC 9254 (Genoa, 24-core)
CPU #2
AMD EPYC 9254 (different physical CPU, different seller)
RAM
SK Hynix DDR5 ECC RDIMM
64GB
HMCG94MEBRA103N
PC5-4800B
PSU #1
Seasonic PRIME TX-1600 Noctua Edition
PSU #2
Another TX-1600 unit with different cables
Symptom
The symptom is IDENTICAL across both motherboards and both CPUs.
BMC works perfectly:
- IPMI accessible
- sensor page accessible
- firmware updates work
- login works
- UID button works
- no fan spin
- no Dr.Debug code
- no VGA output
- no POST
- no power-on
IPMI Power On gives:
Performing Power Action.. Please Wait
Retrying...please wait. Retries Left : 2
Retrying...please wait. Retries Left : 1
Performing power action failed.
Host never enters ON state.
What I tested
Motherboards
- GENOAD8X-2T/BCM
- TURIND8-2L2T
CPUs
- EPYC 9254 #1
- EPYC 9254 #2
Power supplies
- TX-1600 #1
- TX-1600 #2
Memory
- full population
- 1 DIMM in A1
- no DIMMs installed at all
Build configuration
- inside chassis
- outside chassis on cardboard
- minimal config
Cooling
- Arctic SP5 cooler installed
- cooler removed
- tested without cooler for power sequencing
CPU installation
- CPU always kept in carrier
- followed SP5 rail/carrier procedure
- inspected sockets carefully
- no obvious bent pins
- used torque screwdriver
- tested around 1.5 Nm
- BMC updated
- BIOS updated on GENOAD8X
TURIND8 sensor page
Only standby rails exist:
- 3.3VSB present
- 5VSB present
- VCORE disabled
- VSOC disabled
- VOLT_12V disabled
- TEMP_CPU disabled
- POWER_CPU disabled
- all fan sensors disabled
- all DDR5 sensors disabled
System Inventory
Processor Info:
Information Not Available
No CPU information shown.
LEDs
- Standby power LED active
- BMC heartbeat active
- Dr.Debug completely dark
Why I'm confused
At this point I have changed:
- motherboard
- CPU
- PSU
That makes me think this is either:
- a very specific SP5 platform issue
- some CPU presence / host power sequencing failure
- something obvious that I'm completely blind to
- two different SP5 boards
- two different EPYC 9254s
- two PSU units
Has anyone seen:
- Processor Info unavailable
- all CPU sensors disabled
- BMC fully alive
- host power retries and fails
- Dr.Debug completely dark
Any ideas appreciated.
