RTX PRO 6000 Blackwell Dell 1028:204B: identical GSP-FMC 0xaff/0xb on EPYC Gen5 and X79, drivers 580/595/610

kepler-nvidia-bug-report-sanitized.log.gz (1.2 MB)

x79-nvidia-bug-report-sanitized.log.gz (122.3 KB)

I am testing a Dell OEM NVIDIA RTX PRO 6000 Blackwell Workstation Edition (10de:2bb1, subsystem 1028:204b). Linux detects the PCIe device, binds the open NVIDIA kernel module, and the card provides UEFI display output, but GSP initialization fails and nvidia-smi reports No devices were found.

I reproduced this on two very different systems:

  • Gigabyte MZ73-LM2, dual AMD EPYC, Ubuntu 26.04, kernel 7.0.0-28, using the 580, 595 and 610 driver branches; attached report uses 610.43.02.

  • ASRock X79 Extreme6, Xeon E5-1620 v2, Ubuntu 24.04.4, kernel 6.8.0-136, driver 580.173.02.

Representative errors:

Fatal GSP-FMC Error: version=0x1, partition=0x1,
error code=0xaff, additional info=0xb

GSP-FMC reported an error while attempting to boot GSP: 0xb
RmInitAdapter failed! (0x62:0x55:...)

In both machines, the GPU itself was powered by a Corsair HX1500i using its native 12V-2x6 cable. I also tried cold power cycles, enabling ReBAR. PCIe enumeration is stable and I have not found accompanying AER or power errors.

/proc/driver/nvidia identifies the correct model, but reports:

Video BIOS: ?? .??.??.??.??
GPU Firmware: N/A

Before purchase, the seller demonstrated the card producing display output in Windows. I also vaguely recall an NVIDIA application running a 3D display and showing card information, although no compute or stress test was performed while I was present.

Does 0xaff / 0xb identify a specific card-side GSP-FMC, firmware, security, or VBIOS failure? Is there another supported diagnostic or firmware-recovery path for this Dell subsystem, or do these results indicate that the card requires service/RMA?

Sanitized nvidia-bug-report.sh logs from both systems are attached.