Updating the NVMe Firmware

The following instructions describe how to update the firmware on an NVMe device using the NVME-CLI command-line interface. The nvme-cli tool is preinstalled on all DGX H100 systems.

  1. List the devices and check the firmware versions.

    $ sudo nvme list
    
    Node           SN               Model                        Namespace Usage                     Format           FW Rev
    -------------- ---------------- ---------------------------- --------- ------------------------- ---------------- --------
    /dev/nvme0n1   S666NE0T395778   SAMSUNG MZ1L21T9HCLS-00A07   1           1.18  TB /   1.92  TB   512   B +  0 B   GDC7302Q
    /dev/nvme1n1   S666NE0T395802   SAMSUNG MZ1L21T9HCLS-00A07   1           1.18  TB /   1.92  TB   512   B +  0 B   GDC7302Q
    /dev/nvme2n1   22L0A031T2N8     KCM6DRUL3T84                 1          27.04  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme3n1   22W0A02AT2N8     KCM6DRUL3T84                 1          27.57  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme4n1   22W0A02HT2N8     KCM6DRUL3T84                 1          26.19  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme5n1   22L0A01WT2N8     KCM6DRUL3T84                 1          25.44  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme6n1   22L0A025T2N8     KCM6DRUL3T84                 1          24.91  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme7n1   22L0A02CT2N8     KCM6DRUL3T84                 1          25.17  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme8n1   22L0A01FT2N8     KCM6DRUL3T84                 1          24.65  GB /   3.84  TB   512   B +  0 B   0107
    /dev/nvme9n1   22L0A01YT2N8     KCM6DRUL3T84                 1          24.37  GB /   3.84  TB   512   B +  0 B   0107
    

    From the output, you can find the device names and firmware versions, such as /dev/nvme0n1 and GDC7302Q.

  2. Download the firmware you want to upgrade for the NVMe device.

    $ nvme fw-download $DeviceName --fw=***(specify the firmware file for upgrade)
    
  3. Commit and activate the downloaded firmware immediately without reset by setting the action argument to 3.

    $ nvme fw-commit $DeviceName --action=3
    
  4. Verify that the correct firmware version is updated.

    $ nvme id-ctrl $DeviceName