QNN LPAI Predict

Predict responses of a QNN context binary for the LPAI backend

Since R2025b

• 

Libraries:
Embedded Coder Support Package for Qualcomm Hexagon Processors / Hexagon / QNN

Description

The QNN LPAI Predict block predicts responses of a deep learning network represented as a QNN context binary for the LPAI backend of Qualcomm^® AI Direct Engine, based on the given input data.

To add the block to your Simulink model, open the model (for example, myQNNModel), and enter this command at the MATLAB prompt:

add_block("mwqnnlib/QNN LPAI Predict","myQNNModel/QNN LPAI Predict")

The block allows you to select a QNN model as a compiled shared object (.so) for running on x86-based host. For the target, you can select a QNN context binary file (.bin) that is optimized to run on LPAI backend.

The code generated using this block can be deployed to Qualcomm Android Board that is available under the Hardware board parameter in Configuration Parameters.

The block also provides the option to dequantize outputs to single-precision, if required.

Ports

Input

Port_1 — Input signal to predict response

n–D array

The input tensor used for inference with the selected QNN model, represented as an n-D array, in accordance with the Input layer size parameter of the QNN model.

The QNN LPAI Predict block supports a multiple-input multiple-output tensor with a maximum of 4 dimensions, but the batch size must always be 1. For example, if the input layer of the original deep learning network is 128-by-128-by-3, the input dimension should be either 128-by-128-by-3 or 1-by-128-by-128-by-3.

If the leading dimensions are 1 (singleton dimensions), these dimensions can often be removed without affecting compatibility. For example, if the input layer of an AI model expects an input size of 1-by-1-by-128-by-3, the input can be provided as 1-by-1-by-128-by-3 or simply 128-by-3. This is because dimensions of size 1 can be broadcast to match the expected shape.

The QNN LPAI Predict block accepts either floating-point input or fixed-point input. The input datatype must be as per the QNN network's input layer datatype. Additionally, the input can be floating-point even for quantized QNN network.

Data Types: single | half | int8 | int16 | int32 | uint8 | uint16 | uint32

Output

Port_2 — Output signal after inference

n–D array

The output tensor used for inference with the selected QNN model, represented as an n-D array, in accordance with the QNN output layer. The output datatypes match the QNN network's output layers' datatypes. If you select Dequantize Ouptut, the output is always single.

Data Types: single | half | int8 | int16 | int32 | uint8 | uint16 | uint32

Parameters

Host(x86) QNN-model file — QNN model on host (x-86) to predict responses

<filename>.dll

Click Browse and select the QNN model (in .dll format) to perform inference. For details on creating an QNN model to run on device processors like LPAI, refer to Qualcomm AI Engine Direct SDK documentation.

Note

This parameter does not appear if the host operating system is Linux^® (it appears only for Windows host). In the case of Linux, the block uses the same context binary file (which you select using the Context binary file parameter) as the host(x86) QNN-model file also.

Context binary file — QNN context binary file on LPAI backend

<filename>.bin

Click Browse and select the QNN context binary file (.bin) on the target to perform inference on the host. For details on creating an binary file to run on device processors like LPAI, refer to Qualcomm AI Engine Direct SDK documentation.

Dequantize output — Use output dequantization to predict response

off (default) | on

Select the checkbox to dequantize the block's output. Enabling this option results in output data type always being single, irrespective of the deep learning neural network's output layer data type.

Extended Capabilities

C/C++ Code Generation
Generate C and C++ code using Simulink® Coder™.

Version History

Introduced in R2025b