Hailo-10H PCIe – HAILO_NOT_IMPLEMENTED Error When Running Inference on Windows

Usman_ahmed · October 6, 2025, 3:46pm

Hi,
I am attempting to run inference using the pyHailoRT API with a Hailo-10H PCIe device on Windows 11. The device is detected correctly with hailortcli scan, but when I call VDevice.configure() in Python to load a .hef file, it fails with:

hailo_platform.pyhailort.pyhailort.HailoRTException: libhailort failed with error: 7 (HAILO_NOT_IMPLEMENTED)

Environment:

OS: Windows 11 Pro 64-bit
Device: Hailo-10H (PCIe, address 0000:02:00.0)
HailoRT version: [insert exact version shown in hailortcli --version]
pyHailoRT installed from: C:\Program Files\HailoRT\python wheel
Python version: 3.11
Model: yolov11l.hef (from Model Zoo)

Steps to Reproduce:

Installed HailoRT MSI (checked pyHailoRT option).
Confirmed the card appears in hailortcli scan.

Ran a Python script that creates a VDevice and configures the HEF via:

configure_params = ConfigureParams.create_from_hef(hef=hef, interface=HailoStreamInterface.PCIe)
network_groups = target.configure(hef, configure_params)

The above call fails with error code 7 (HAILO_NOT_IMPLEMENTED).

Expected Behavior:
The HEF should be configured successfully, allowing inference through PCIe on Windows.

Actual Behavior:
The device is detected but configuration fails with HAILO_NOT_IMPLEMENTED.

Questions:

Is PCIe inference with pyHailoRT currently supported on Windows for Hailo-10H?
If not, is there a planned release adding PCIe support, or is Linux required for hardware inference?
Can simulation mode be used to validate the pipeline on Windows until full PCIe support is available?

Thank you for your assistance.

nina-vilela · October 6, 2025, 4:02pm

Hi @Usman_ahmed,

Windows inference on Hailo-10H PCIe is supported with your driver version, but the API you used is old and only works with Hailo-8.

For Hailo-10H, the supported option is the asynchronous API, which also provides better performance. Since you’re using YOLOv11, you can use this example:

It uses the new API and works on Windows with Hailo-10h, but for camera streams you need to change this line:

github.com/hailo-ai/Hailo-Application-Code-Examples

runtime/hailo-8/python/common/toolbox.py

main


      
                  exit(1)
          
              if not resolution:
                  CAMERA_CAP_WIDTH, CAMERA_CAP_HEIGHT = get_camera_native_resolution()
              else:
                  CAMERA_CAP_WIDTH, CAMERA_CAP_HEIGHT = CAMERA_RESOLUTION_MAP.get(resolution, (640, 480))  # fallback to SD
          
              cap = cv2.VideoCapture(CAMERA_INDEX)
              cap.set(cv2.CAP_PROP_FRAME_WIDTH, CAMERA_CAP_WIDTH)
              cap.set(cv2.CAP_PROP_FRAME_HEIGHT, CAMERA_CAP_HEIGHT)
              cap.set(cv2.CAP_PROP_FOURCC, cv2.VideoWriter_fourcc(*'YUYV'))
          
          elif any(input_path.lower().endswith(suffix) for suffix in ['.mp4', '.avi', '.mov', '.mkv']):
              if not os.path.exists(input_path):
                  logger.error(f"File not found: {input_path}")
                  sys.exit(1)
              cap = cv2.VideoCapture(input_path)
          else:
              images = load_images_opencv(input_path)
              try:
                  validate_images(images, batch_size)

to:
cap.set(cv2.CAP_PROP_FOURCC, cv2.VideoWriter_fourcc(*‘MJPG’))

Usman_ahmed · October 6, 2025, 8:04pm

Thank you for the details answer issue is solved.