Qualcomm AI Hub example NPU error

vasucp1207 · January 2, 2026, 9:22am

I’m trying to run the face_det_lite example from the AI Model Hub following the docs.

Passing the —use-npu flag results in QNN Delegate error.


python3 face_detection.py --use-npu


/prj/qct/webtech_hyd18/mlg_user_admin/qaisw_source_repo/qaisw_repo_point_release/snpe_src/avante-tools/prebuilt/dsp/hexagon-sdk-4.2.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:38:dummy call to rpcmem_init, rpcmem APIs will be used from libxdsprpc

ERROR: [QNN Delegate] Failed to create device_handle for Backend ID 6, error=14001

Traceback (most recent call last):

File "/home/ubuntu/qai/face_detection.py", line 22, in <module>

interpreter = Interpreter(

^^^^^^^^^^^^

File "/home/ubuntu/qnn_venv/lib/python3.12/site-packages/ai_edge_litert/interpreter.py", line 536, in __init__

self._interpreter.ModifyGraphWithDelegate(

RuntimeError: Restored original execution plan after delegate application failure.

System Info:

Os image link


PRETTY_NAME="Ubuntu 24.04.3 LTS"

NAME="Ubuntu"

VERSION_ID="24.04"

VERSION="24.04.3 LTS (Noble Numbat)"

VERSION_CODENAME=noble

ID=ubuntu

ID_LIKE=debian

HOME_URL="https://www.ubuntu.com/"

SUPPORT_URL="https://help.ubuntu.com/"

BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"

PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"

UBUNTU_CODENAME=noble

LOGO=ubuntu-logo

qnn-platform-validator test is also failing


qnn-platform-validator --backend dsp --testBackend


PF_VALIDATOR: DEBUG: Calling PlatformValidator->setBackend

PF_VALIDATOR: DEBUG: Calling PlatformValidator->isBackendHardwarePresent

PF_VALIDATOR: DEBUG: Calling PlatformValidator->isBackendAvailable

PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libc.so.6

PF_VALIDATOR: DEBUG: dlOpen successfull for library : libc.so.6

PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libcdsprpc.so

PF_VALIDATOR: DEBUG: dlOpen successfull for library : libcdsprpc.so

Backend DSP Prerequisites: Present.

PF_VALIDATOR: DEBUG: Calling PlatformValidator->backendCheck

PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libc.so.6

PF_VALIDATOR: DEBUG: dlOpen successfull for library : libc.so.6

PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libcdsprpc.so

PF_VALIDATOR: DEBUG: dlOpen successfull for library : libcdsprpc.so

PF_VALIDATOR: DEBUG: Starting calculator test

PF_VALIDATOR: DEBUG: Loading sample stub: libQnnHtpV68CalculatorStub.so

PF_VALIDATOR: DEBUG: Successfully loaded DSP library - 'libQnnHtpV68CalculatorStub.so'. Setting up pointers.

Error opening session with unsigned PD

PF_VALIDATOR: ERROR: -1 . Error while executing the sum function.

PF_VALIDATOR: ERROR: Please use testsig if using unsigned images.

PF_VALIDATOR: ERROR: Also make sure ADSP_LIBRARY_PATH points to directory containing skels.

Unit Test on the backend DSP: Failed.

QNN is NOT supported for backend DSP on the device.

*********** Results Summary ***********

Backend = DSP

{

Backend Hardware : Supported

Backend Libraries : Found

Library Version : Not Queried

Core Version : Not Queried

Unit Test : Failed

}

Error in saving the results ⏎

kinkin · January 4, 2026, 7:19am

May I ask whether you are using Ubuntu Server or Ubuntu Desktop?
Have you also used root privileges?
We recommend testing with the latest version; you can refer to the following steps to upgrade.

If the issue persists, please share the ~/aihub-demo directory from your development board.

vasucp1207 · January 5, 2026, 4:31am

I am using Ubuntu desktop and using its latest version. I tried with root privileges but got the same errors no difference.

I am wondering what can be the reason for the error.
Error opening session with unsigned PD, I see a lot of people going through this issue but not got a final answer to solve this

kinkin · January 5, 2026, 11:06am

The issues identified in the Ubuntu Desktop image will be optimized in a future release.
For the time being, we recommend that you use the Ubuntu Server image for your AI-development activities, provided this alternative is acceptable to you.

vasucp1207 · January 5, 2026, 12:36pm

I just checked with the Ubuntu server image and there is no difference at all. I got the same error as the desktop version.

kinkin · January 6, 2026, 2:30am

We have verified that the Ubuntu Server image functions correctly in our environment.
Would you kindly share the ~/aihub-demo directory from your development board?

vasucp1207 · January 6, 2026, 4:27am

I followed this demo and there is only one file face_detection.py.
I transferred the qnn libs to the board and set LD_LIBRARY_PATH and ADSP_LIBRARY_PATH
After running the python3 face_detection.py --use-npu, I got

<W> Initializing HtpProvider
 <W> Specified config SOC, ignoring on real target
 <E> Fail to get effective domain id from rpc with DeviceId 0 coreId 0 pdId 0
 <E> error in creation of transport instance
 <W> Failed to create transport instance: 1002
 <E> Failed to create transport for device, error: 1002
 <E> Failed to load skel, error: 1002
 <E> Transport layer setup failed: 14001
 <E> Failed to parse default platform info: 14001                                                                    <E> Failed to load default platform info: 14001
 <E> Failed to parse platform config: 14001
Traceback (most recent call last):
  File "/home/ubuntu/litert-demo/face_detection.py", line 22, in <module>                                               interpreter = Interpreter(
                  ^^^^^^^^^^^^
  File "/home/ubuntu/litert-demo/.venv/lib/python3.12/site-packages/ai_edge_litert/interpreter.py", line 536, in __init__                                                                                                                   self._interpreter.ModifyGraphWithDelegate(
RuntimeError: Restored original execution plan after delegate application failure.

Also the qnn-platform-validator dsp test also keep failing.

--backend dsp --testBackend
PF_VALIDATOR: DEBUG: Calling PlatformValidator->setBackend
PF_VALIDATOR: DEBUG: Calling PlatformValidator->isBackendHardwarePresent
PF_VALIDATOR: DEBUG: Calling PlatformValidator->isBackendAvailable
PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libc.so.6
PF_VALIDATOR: DEBUG: dlOpen successfull for library : libc.so.6
PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libcdsprpc.so
PF_VALIDATOR: DEBUG: dlOpen successfull for library : libcdsprpc.so
Backend DSP Prerequisites: Present.
PF_VALIDATOR: DEBUG: Calling PlatformValidator->backendCheck
PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libc.so.6
PF_VALIDATOR: DEBUG: dlOpen successfull for library : libc.so.6
PF_VALIDATOR: DEBUG: Should be able to access atleast one of libraries from : libcdsprpc.so
PF_VALIDATOR: DEBUG: dlOpen successfull for library : libcdsprpc.so
PF_VALIDATOR: DEBUG: Starting calculator test
PF_VALIDATOR: DEBUG: Loading sample stub: libQnnHtpV68CalculatorStub.so
PF_VALIDATOR: DEBUG: Successfully loaded DSP library - 'libQnnHtpV68CalculatorStub.so'.  Setting up pointers.       Error opening session with unsigned PD
Unable to destroy the handlePF_VALIDATOR: ERROR: -6 . Error while executing the sum function.
PF_VALIDATOR: ERROR: Please use testsig if using unsigned images.
PF_VALIDATOR: ERROR: Also make sure ADSP_LIBRARY_PATH points to directory containing skels.                         Unit Test on the backend DSP: Failed.
QNN is NOT supported for backend DSP on the device.

kinkin · January 6, 2026, 7:51am

May I ask whether you performed the test using the method below?
If so, please share the ~/aihub-demo directory, as we will use it for testing and to gather further information.
If not, kindly refer the method below first to confirm that python3 face_detection.py --use-npu executes successfully.

kinkin · January 8, 2026, 11:14am

Update:
If the issue persists, please upload the ~/aihub-npu directory.

vasucp1207 · January 8, 2026, 11:24am

I found out the solution, this is due to the qnn libs version mismatch and the wrong version of qairt sdk. For now it’s working for me.

kinkin · January 9, 2026, 1:48am

Great news! Thank you for your response.

Rashedkoutayni · January 14, 2026, 10:35am

@vasucp1207 How did you solve the version mismatch issue please? I have the same problem I guess.

vasucp1207 · January 15, 2026, 5:04am

I installed the v2.37.1.250807 qairt sdk and removed all the dependencies from the old one. Kept the complete sdk on your board and set the env variables.

export QNN_SDK_ROOT=<path>/2.37.1.250807
export ADSP_LIBRARY_PATH=$QNN_SDK_ROOT/lib/hexagon-v$DSP_ARCH/unsigned
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$QNN_SDK_ROOT/lib/aarch64-oe-linux-gcc11.2
export PRODUCT_SOC=6490 DSP_ARCH=68

# test for the dsp backend(make sure qnn-platform-validator points to right location)
qnn-platform-validator --backend dsp --testBackend