NVIDIA Field Diagnostics Test (FDIAG) Tool for Validating H100/H200

NVIDIA Field Diagnostics Test (FDIAG) Tool for Validating H100/H200


FDIAG
NVIDIA’s Field Diagnostics Test Tool (FDIAG)is an offline hardware test tool for GPU health test. Mainly used in RMA workflow and tool is not provided to public for download. They are distributed through OEMs or directly from NVIDIA support as part of service. It covers log generation, FDIAG execution, and log collection for RMA or system verification.

Prerequisites

  • Portable bootable USB (Service OS)
  • USB hub
  • USB data stick for log collection
  • Keyboard and monitor (can be removed after FDIAG starts)

Step 1: Boot into Service OS

  1. Connect the following:

        Bootable Ubuntu USB stick
        USB hub
        USB data stick for Collecting Log
  2. Power on or restart the system.
  3. Press F11 to access the Boot Menu.
  4. Select the portable USB stick and boot into Service OS (Ubuntu). If USB Stick is not visible, go to BIOS and change the boot settings UEFI to dual mode. After Complete the test change back to UEFI mode.
  5. Log in using the following credentials:
    Username: service
    Password: 1234

 

Step 2: Initial GPU & System Verification

Perform the following checks before starting FDIAG.

1. Verify GPU Detection

nvidia-smi

2. Create GPU Serial Number List

nvidia-smi -q | grep -i serial >> nvidiaserials.txt

3. View GPU Serial Numbers

vi nvidiaserials.txt

 

4. Verify GPU VBIOS Firmware

nvidia-smi -q | grep -i Vbios

5. Verify BIOS and BMC Versions

sudo ./IPMICFG-Linux.x86_64 -summary

Step 3: Running NVIDIA Field Diagnostics (FDIAG)

  1. Change directory to the H100 folder:

Type - cd H100 to change into H100 folder. You can run ls to confirm that the Python script is available.

  1. Start FDIAG:

Type - python3 run.py

  1. Select the system type:

Option 1 – Intel system

Option 2 – AMD system

  1. Select the test mode:

Option 4 – RMA

Option 5 – System Validation

  1. Wait until the DGX FIELD DIAGNOSTIC messages appear

 

  1. Once FDIAG has started we can safely remove Keyboard and monitor

Important Notes

  • H100 / H200 FDIAG runtime: Approximately 75–90 minutes
  • A100 FDIAG runtime: Approximately 5 hours
  • Always start the sequence as: 
              For Intel Systems start 1 to 4/.5
              For AMD Systems start 2 to 4/5

Once FDIAG begins, you cannot return to generate logs. PASS or FAIL Screen will appear once test finished.
FAIL Screen Image is Attached

 

Step 4: Log Collection After FDIAG Completion

1. Verify All Drives
           Type - lsblk

2. Identify Data USB
             Typically identified as sdb/sdd

3. Mount the USB Partition
         Type -sudo mount /dev/sdb1 /media/usb/

4. Copy or Move Logs
Copy or move the System Serial Number folder to the USB:
       Type - sudo cp -r <System_Serial_Number_Folder> /media/usb OR Type - sudo mv <System_Serial_Number_Folder> /media/usb

5. Synchronize Data
     Type - sync

6. Verify USB Contents
        Type - cd /media/usb
         ls (or) dir -i

Step 5: USB Removal and Upload

  1. Unmount the USB safely: 
    Type - sudo umount /media/usb/

  1. Remove the USB data stick.
  2. Connect it to your laptop.
  3. Upload the collected folder to the shared drive / Box folder.
  4. Shut down Ubuntu OS and remove all USB devices.