# UDO Tutorial

Overview

This tutorial describes the steps needed to create a UDO
package and execute the Inception V3 model using the package.
The Softmax operation has been chosen in this tutorial to
demonstrate the implementation of a UDO with Qualcomm® Neural Processing SDK.

The Qualcomm® Neural Processing SDK provides the resources for this example under

- $SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax

Information on UDO in general is available at [UDO Overview](https://docs.qualcomm.com/doc/80-63442-2/topic/udo_overview.html).

Information on running the Inception V3 network without UDO is
available at [Inception V3 Tutorial](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3.html).

Prerequisites

The following tutorial assumes that general [Qualcomm (R) Neural Processing SDK
setup](https://docs.qualcomm.com/doc/80-63442-2/topic/setup.html) has been followed to support
SDK environment, TensorFlow environment, and desired platform
dependencies. Please ensure that your hexagon sdk is named in the format “hexagon-sdk-<version name>” and ensure
that the hexagon sdk has the path for the tools directory in format of “hexagon-sdk-<version name>/tools/HEXAGON\_Tools/<tools name>/”

Additionally, we need an extracted Qualcomm® AI Direct SDK (no
need of Qualcomm® AI Direct SDK setup) for generating the skeleton code and
building the libraries. For Qualcomm® AI Direct SDK details, refer to the Qualcomm® AI Direct SDK
documentation at `$QNN_SDK_ROOT/docs/QNN/index.html` page, where
`QNN_SDK_ROOT` is the location of the Qualcomm® AI Direct SDK installation.
Set the `$QNN_SDK_ROOT` to the unzipped Qualcomm® AI Direct SDK location. This has to be performed
after running the envsetup.sh script mentioned in [SNPE Setup](https://docs.qualcomm.com/doc/80-63442-2/topic/setup.html#environment-setup). The
steps listed in this tutorial use the Tensorflow model in the
form of inception\_v3\_2016\_08\_28\_frozen.pb. For details on
acquiring the Inception V3 model visit [Tutorials
Setup](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_setup.html#getting-inception-v3).

Introduction

Here are the steps to develop and run a UDO

1. [Package Generation](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#step-1-package-generation)
2. [Framework Model Conversion to a DLC](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#step-2-framework-model-conversion-to-a-dlc)
3. [Package Implementation](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#step-3-package-implementations)
4. [Package Compilation](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#step-4-package-compilation)
5. [Model Execution](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#model-execution)

Steps 1-4 are run offline on the x86 host and are necessary for
execution in step 5. Step 5 provides information on execution
using the Qualcomm® Neural Processing SDK command-line executable **snpe-net-run**.
Optionally, the user can perform steps 1-4 automatically using
the provided [setup script](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#setup-script).

Make sure to run following commands for setting up the tensorflow paths and hexagon path

export TENSORFLOW_HOME=<path to tensorflow framework version 2.3.0>
    export TENSORFLOW_ROOT=$TENSORFLOW_HOME/focal/py3
    export PYTHONPATH=$TENSORFLOW_ROOT/distribute:$TENSORFLOW_ROOT/dependencies/python:$PYTHONPATH
    
    # hexagon sdk and tools version depend on dsp aarch, below version is for v68 aarch
    export HEXAGON_SDK_ROOT=<path to Hexagon SDK 4.2.0>
    export HEXAGON_TOOLS_ROOT=$HEXAGON_SDK_ROOT/tools/HEXAGON_Tools/8.4.09
    Copy to clipboard

Step 1: Package Generation

Generating the SoftmaxUdoPackage requires the
**snpe-udo-package-generator** tool and the provided UDO
plugin: Softmax\_Htp.json. The plugin is located under
$SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/config. More
information about creating a UDO plugin can be found
[here](https://docs.qualcomm.com/doc/80-63442-2/topic/udo_operator_definition.html#the-udo-configuration-specification).

Generate the SoftmaxUdoPackage using the following:

export SNPE_UDO_ROOT=$SNPE_ROOT/share/SNPE/SnpeUdo
    export QNN_SDK_ROOT=<path to Qualcomm® AI Direct SDK>
    snpe-udo-package-generator -p $SNPE_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/config/Softmax_Htp.json -o $SNPE_ROOT/examples/Models/InceptionV3/
    Copy to clipboard

This command creates the Softmax based package at
$SNPE\_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage.
For more information on the snpe-udo-package-generator tool
visit [here](https://docs.qualcomm.com/doc/80-63442-2/topic/creating_udo_package.html).

Step 2: Framework model Conversion to a DLC

Converting the Tensorflow Inception V3 model to DLC requires
the [snpe-tensorflow-to-dlc](https://docs.qualcomm.com/doc/80-63442-2/topic/tools.html#snpe-tensorflow-to-dlc) tool.
The snpe-tensorflow-to-dlc
tool consumes the same Softmax\_Htp.json used in package generation
via the –udo command line option. In this step,
&lt;INCEPTION\_V3\_PATH&gt; refers to the path to the inception\_v3 pb
file. For example, after running the setup\_inceptionv3\_snpe.py
script &lt;INCEPTION\_V3\_PATH&gt; is
$SNPE\_ROOT/examples/Models/InceptionV3/tensorflow.

Convert Inception V3 with the following:

snpe-tensorflow-to-dlc --input_network <INCEPTION_V3_PATH>/inception_v3_2016_08_28_frozen.pb --input_dim 'input' 1,299,299,3 --out_node InceptionV3/Predictions/Reshape_1 --output_path $SNPE_ROOT/examples/Models/InceptionV3/dlc/inception_v3_udo.dlc --udo $SNPE_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/config/Softmax_Htp.json
    Copy to clipboard

This will generate a DLC named inception\_v3\_udo.dlc containing
the Softmax as UDO at $SNPE\_ROOT/examples/Models/InceptionV3/dlc.

Step 3: Package Implementations

The generated package creates the skeleton of the operation
implementation, which must be filled by the user to create a
functional UDO. The rest of the code scaffolding for
compatibility with Qualcomm® Neural Processing SDK is provided by the
**snpe-udo-package-generator**. The UDO implementations for this tutorial are provided under
$SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src.

**CPU Implementations (Android and x86)**

The file in the package that needs to be implemented for CPU is

- SoftmaxUdoPackage/jni/src/CPU/src/ops/Softmax.cpp

The provided example implementation is present at the location

- $SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/CPU/Softmax.cpp

Copy the provided implementation to the package:

cp -f $SNPE_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/CPU/Softmax.cpp $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/jni/src/CPU/src/ops/
    Copy to clipboard

Optionally, the user can provide their own implementations in
the package.

**GPU Implementations (Android)**

The file in the package that needs to be implemented for GPU is

- SoftmaxUdoPackage/jni/src/GPU/src/ops/Softmax.cpp

The provided example implementation is present at the location

- $SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/GPU/Softmax.cpp

Copy the provided implementation to the package:

cp -f $SNPE_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/GPU/Softmax.cpp $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/jni/src/GPU/src/ops/
    Copy to clipboard

Optionally, the user can provide their own implementations in
the package.

**DSP Implementations for V65 and V66**

Similar to all other Qualcomm® Neural Processing SDK runtimes, a registration library and
an implementation library are required to run inference of a
network with UDO layers on Qualcomm® Neural Processing SDK DSP. The registration library
will run on CPU, and specifies the DSP implementation library
of the UDO. Refer [Implementing a UDO for DSP V65 and
V66](https://docs.qualcomm.com/doc/80-63442-2/topic/compiling_udo_package.html#implementing-a-udo-for-dsp-v65-and-v66)
for more information on implementing UDO for DSP V65 and V66
runtimes.

The file in the package that need to be implemented for DSP V65
and V66 are

- SoftmaxUdoPackage/jni/src/DSP/Softmax.cpp

The provided example implementation is present at the location

- $SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/DSP/Softmax.cpp

Copy the provided implementations to the package:

cp -f $SNPE_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/DSP/Softmax.cpp $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/jni/src/DSP/src/ops
    Copy to clipboard

Optionally, the user can provide their own implementations in
the package.

**DSP Implementations for V68 and later**

Refer [Implementing a UDO for DSP V68 or
later](https://docs.qualcomm.com/doc/80-63442-2/topic/compiling_udo_package.html#implementing-a-udo-for-dsp-v68-or-later)
for more information on implementing UDO for DSP V68 or later
runtimes. The directory paths and locations in this example are
specific to DSP V68. For later runtimes, please replace
**DSP\_V68** with the corresponding DSP architecture (for
example, **DSP\_V69**) in the paths.

The file in the package that needs to be implemented for DSP
V68 and later is

- SoftmaxUdoPackage/jni/src/DSP\_V68/src/ops/Softmax.cpp

The provided example implementation is present at the location

- $SNPE\_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/HTP/Softmax.cpp

Copy the provided implementations to the package:

cp -f $SNPE_ROOT/examples/SNPE/NativeCpp/UdoExample/Softmax/src/HTP/Softmax.cpp $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/jni/src/DSP_V68/src/ops
    Copy to clipboard

Optionally, the user can provide their own implementations in
the package.

Step 4: Package Compilation

**x86 Host Compilation**

Compiling on x86 host uses the make build system. Compile the
CPU implementations with the following:

cd $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage
    make cpu_x86
    Copy to clipboard

The expected artifacts after compiling for CPU on x86 host are

- SoftmaxUdoPackage/libs/x86-64\_linux\_clang/libUdoSoftmaxUdoPackageImplCpu.so
- SoftmaxUdoPackage/libs/x86-64\_linux\_clang/libUdoSoftmaxUdoPackageReg.so

**Android CPU Runtime Compilation**

Compilation for the CPU runtime on Android uses Android NDK.
The ANDROID\_NDK\_ROOT environment variable must be set to the
directory containing ndk-build in order to compile the package.

export ANDROID_NDK_ROOT=<path_to_android_ndk>
    Copy to clipboard

It is suggested to add ANDROID\_NDK\_ROOT to the PATH environment
variable to access ndk-build.

export PATH=$ANDROID_NDK_ROOT:$PATH
    Copy to clipboard

Once the ANDROID\_NDK\_ROOT is part of PATH, compile the package
for Android CPU target:

cd $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage
    make cpu_android
    Copy to clipboard

The expected artifacts after compiling for Android CPU are

- SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageImplCpu.so
- SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageReg.so
- SoftmaxUdoPackage/libs/arm64-v8a/libc++\_shared.so

**Hexagon DSP Runtime Compilation**

Compilation for the DSP runtime makes use of the make system.
In order to build the implementation libraries for DSP V65 and
V66 runtimes, Hexagon-SDK needs to be installed and set up. For
details, follow the setup instructions on
`$HEXAGON_SDK_ROOT/docs/readme.html` page, where
`HEXAGON_SDK_ROOT` is the location of your Hexagon-SDK
installation. Information for compiling a UDO for DSP is
available at [Compiling UDO for
DSP](https://docs.qualcomm.com/doc/80-63442-2/topic/compiling_udo_package.html#compiling-a-udo-for-dsp-v65-and-v66-on-device).

Setup Script

The Qualcomm® Neural Processing SDK provides an option to automatically perform steps
of DLC conversion, package generation, package implementation,
and package compilation for UDO as outlined in steps 1-4 above.
The option is an extension of the [Inception V3 setup
script](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_setup.html#getting-inception-v3). To
enable Inception V3 setup for UDO, run the script with the
**–udo** or **-u** option.

usage: $SNPE_ROOT/models/examples/Models/InceptionV3/scripts/setup_inceptionv3_snpe.py [-h] -a ASSETS_DIR [-d] [-r RUNTIME] [-u] [-l [HTP_SOC]]
    
    Prepares the InceptionV3 assets for tutorial examples.
    
    required arguments:
      -a ASSETS_DIR, --assets_dir ASSETS_DIR
                            directory containing the InceptionV3 assets
    
    optional arguments:
      -d, --download        Download InceptionV3 assets to InceptionV3 example
                            directory
      -r RUNTIME, --runtime RUNTIME
                            Choose a runtime to set up tutorial for. Choices: cpu,
                            gpu, dsp, aip, all. 'all' option is only supported
                            with --udo flag
      -u, --udo             Generate and compile a user-defined operation package
                            to be used with InceptionV3. Softmax is simulated as
                            a UDO for this script.
      -l [HTP_SOC], --htp_soc [HTP_SOC]
                            Specify SOC target for generating HTP Offline Cache.
                            For example: "--htp_soc sm8450" for Snapdragon 8 Gen 1,
                            default value is sm8750.
    Copy to clipboard

The –udo extension is compatible with options normally used by
the setup\_inceptionv3\_snpe.py script. When the –udo option is
enabled, the -r or –runtime option controls the runtime for
the package implementation and compilation. Additionally, the
–udo option supports use of an ‘all’ runtime option to create
and compile the SoftmaxUdoPackage for the CPU, GPU, and
DSP/AIP runtimes. Selecting the ‘aip’ or ‘dsp’ runtime options
additionally compiles x86 libraries in order to quantize the
model. Selecting the ‘cpu’ runtime option compiles for both x86
and Android targets. Compilation for Android target will be
skipped if ANDROID\_NDK\_ROOT is not set. If no runtime option is
provided, the package is compiled for the CPU runtime.

The command to use the setup script for UDO is:

python3 $SNPE_ROOT/examples/Models/InceptionV3/scripts/setup_inceptionv3_snpe.py -a ~/tmpdir -d -u -r <runtime_of_choice>
    Copy to clipboard

For instance, to create and compile a package containing the
libraries for all runtimes run:

python3 $SNPE_ROOT/examples/Models/InceptionV3/scripts/setup_inceptionv3_snpe.py -a ~/tmpdir -d -u -r all
    Copy to clipboard

For DSP\_V68 or later, use the -l optional argument of the setup script to specify the htp soc. Or follow the steps [here](https://docs.qualcomm.com/doc/80-63442-2/topic/compiling_udo_package.html#implementing-a-udo-for-dsp-v68-or-later).

This will populate the artifacts in [Step
4](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo.html#step-4-package-compilation).

Model Execution

**Execution using snpe-net-run**

Executing Inception V3 with UDO is largely the same as use of
[snpe-net-run](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3.html#overview)
without UDO.

The Qualcomm® Neural Processing SDK provides Linux and Android binaries of
**snpe-net-run** under

- $SNPE\_ROOT/bin/x86\_64-linux-clang
- $SNPE\_ROOT/bin/aarch64-android
- $SNPE\_ROOT/bin/aarch64-oe-linux-gcc8.2
- $SNPE\_ROOT/bin/aarch64-oe-linux-gcc9.3
- $SNPE\_ROOT/bin/aarch64-ubuntu-gcc9.4
- $SNPE\_ROOT/bin/aarch64-oe-linux-gcc11.2

For UDO, snpe-net-run consumes the registration library through
the –udo\_package\_path option. LD\_LIBRARY\_PATH must also be
updated to include the runtime-specific artifacts generated
from package compilation.

**x86 Host Execution**

To execute the network on x86 host, run:

cd $SNPE_ROOT/examples/Models/InceptionV3
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/libs/x86-64_linux_clang/
    snpe-net-run --container dlc/inception_v3_udo.dlc --input_list data/cropped/raw_list.txt --udo_package_path SoftmaxUdoPackage/libs/x86-64_linux_clang/libUdoSoftmaxUdoPackageReg.so
    Copy to clipboard

**Android Target Execution**

The tutorial for execution on Android targets will use the
arm64-v8a architecture. This portion of the tutorial is generic
to all runtimes (CPU, GPU, DSP). Set SNPE\_TARGET\_DSPARCH
to the DSP architecture of the target Android device.

# architecture: arm64-v8a - compiler: clang - STL: libc++
    export SNPE_TARGET_ARCH=aarch64-android
    export SNPE_TARGET_DSPARCH=hexagon-v68
    Copy to clipboard

Then, push Qualcomm® Neural Processing SDK binaries and libraries to the target device:

adb shell "mkdir -p /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin"
    adb shell "mkdir -p /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib"
    
    adb push $SNPE_ROOT/lib/$SNPE_TARGET_ARCH/*.so \
          /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib
    adb push $SNPE_ROOT/bin/$SNPE_TARGET_ARCH/snpe-net-run \
          /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin
    Copy to clipboard

Next, update environment variables on the target device to
include the Qualcomm® Neural Processing SDK libraries and binaries:

adb shell
    export SNPE_TARGET_ARCH=aarch64-android
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib
    export PATH=$PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin
    Copy to clipboard

Lastly, push the Inception V3 UDO model and input data to the device:

cd $SNPE_ROOT/examples/Models/InceptionV3
    mkdir data/rawfiles && cp data/cropped/*.raw data/rawfiles/
    adb shell "mkdir -p /data/local/tmp/inception_v3_udo"
    adb push data/rawfiles /data/local/tmp/inception_v3_udo/cropped
    adb push data/target_raw_list.txt /data/local/tmp/inception_v3_udo
    adb push dlc/inception_v3_udo.dlc /data/local/tmp/inception_v3_udo
    rm -rf data/rawfiles
    Copy to clipboard

**Android CPU Execution**

Once the model and data have been placed on the device, place
the UDO libraries on the device:

cd $SNPE_ROOT/examples/Models/InceptionV3
    adb shell "mkdir -p /data/local/tmp/inception_v3_udo/cpu"
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageImplCpu.so /data/local/tmp/inception_v3_udo/cpu
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageReg.so /data/local/tmp/inception_v3_udo/cpu
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libc++_shared.so /data/local/tmp/inception_v3_udo/cpu
    Copy to clipboard

Now set required environment variables and run snpe-net-run on device:

adb shell
    cd /data/local/tmp/inception_v3_udo/
    export SNPE_TARGET_ARCH=aarch64-android
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib
    export PATH=$PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin
    export LD_LIBRARY_PATH=/data/local/tmp/inception_v3_udo/cpu/:$LD_LIBRARY_PATH
    snpe-net-run --container inception_v3_udo.dlc --input_list target_raw_list.txt --udo_package_path cpu/libUdoSoftmaxUdoPackageReg.so
    Copy to clipboard

**Android GPU Execution**

The procedure for execution using the GPU runtime is largely
similar to execution procedure for CPU. First, use the
following to place the UDO libraries on device:

cd $SNPE_ROOT/examples/Models/InceptionV3
    adb shell "mkdir -p /data/local/tmp/inception_v3_udo/gpu"
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageImplGpu.so /data/local/tmp/inception_v3_udo/gpu
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageReg.so /data/local/tmp/inception_v3_udo/gpu
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libc++_shared.so /data/local/tmp/inception_v3_udo/gpu
    Copy to clipboard

Now set required environment variables and run snpe-net-run to
device:

adb shell
    cd /data/local/tmp/inception_v3_udo/
    export SNPE_TARGET_ARCH=aarch64-android
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib
    export PATH=$PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin
    export LD_LIBRARY_PATH=/data/local/tmp/inception_v3_udo/gpu/:$LD_LIBRARY_PATH
    snpe-net-run --container inception_v3_udo.dlc --input_list target_raw_list.txt --udo_package_path gpu/libUdoSoftmaxUdoPackageReg.so --use_gpu
    Copy to clipboard

**Hexagon DSP Execution**

The procedure for execution on device for DSP is largely the
same as CPU and GPU. However, the DSP runtime requires
quantized network parameters. While DSP allows unquantized
DLCs, it is generally recommended to quantize DLCs for improved
performance. The tutorial will use a quantized DLC as an
illustrative example. Quantizing the DLC requires the
**snpe-dlc-quantize** tool.

To quantize the DLC for use on DSP:

cd $SNPE_ROOT/examples/Models/InceptionV3/
    snpe-dlc-quantize --input_dlc dlc/inception_v3_udo.dlc --input_list data/cropped/raw_list.txt --udo_package_path SoftmaxUdoPackage/libs/x86-64_linux_clang/libUdoSoftmaxUdoPackageReg.so --output_dlc dlc/inception_v3_udo_quantized.dlc
    Copy to clipboard

For more information on **snpe-dlc-quantize** visit
[quantization](https://docs.qualcomm.com/doc/80-63442-2/topic/quantized_models.html#overview). For
information on UDO-specific quantization visit [Quantizing a
DLC with UDO](https://docs.qualcomm.com/doc/80-63442-2/topic/preparing_model_with_udo.html#quantizing-a-dlc-with-udo).
For information on DSP/AIP runtime visit [DSP
Runtime](https://docs.qualcomm.com/doc/80-63442-2/topic/dsp_runtime.html) or [AIP
Runtime](https://docs.qualcomm.com/doc/80-63442-2/topic/aip_runtime.html).

Now push the quantized model to device:

adb push dlc/inception_v3_udo_quantized.dlc /data/local/tmp/inception_v3_udo
    Copy to clipboard

**Note:** Please refer to [UDO DSP tutorial for Quantized
DLC](https://docs.qualcomm.com/doc/80-63442-2/topic/tutorial_inceptionv3_udo_dsp.html) for executing on the
DSP runtime using quantized dlc.

Before executing on the DSP, push the Qualcomm® Neural Processing SDK libraries for DSP to
device:

adb shell "mkdir -p /data/local/tmp/snpeexample/dsp/lib"
    adb push $SNPE_ROOT/lib/$SNPE_TARGET_DSPARCH/unsigned/*.so /data/local/tmp/snpeexample/dsp/lib
    Copy to clipboard

Now push DSP-specific UDO libraries to device. Depending on DSP
architecture specified in the config, **dsp\_v68** directory can
be **dsp\_v60** or **dsp** (with older Qualcomm® Neural Processing SDK).

cd $SNPE_ROOT/examples/Models/InceptionV3
    adb shell "mkdir -p /data/local/tmp/inception_v3_udo/dsp"
    adb push SoftmaxUdoPackage/libs/dsp_v68/*.so /data/local/tmp/inception_v3_udo/dsp
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageReg.so /data/local/tmp/inception_v3_udo/dsp # Pushes reg lib
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libc++_shared.so /data/local/tmp/inception_v3_udo/dsp
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageImplDsp_AltPrep.so /data/local/tmp/inception_v3_udo/dsp #If online prepare is used
    Copy to clipboard

Then set required environment variables and run snpe-net-run on
device:

adb shell
    cd /data/local/tmp/inception_v3_udo/
    export SNPE_TARGET_ARCH=aarch64-android
    export LD_LIBRARY_PATH=/data/local/tmp/inception_v3_udo/dsp:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib
    export PATH=$PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin
    export ADSP_LIBRARY_PATH="/data/local/tmp/inception_v3_udo/dsp/;/data/local/tmp/snpeexample/dsp/lib;/system/lib/rfsa/adsp;/system/vendor/lib/rfsa/adsp;/dsp"
    snpe-net-run --container inception_v3_udo_quantized.dlc --input_list target_raw_list.txt --udo_package_path dsp/libUdoSoftmaxUdoPackageReg.so --use_dsp
    Copy to clipboard

**AIP Execution**

Because UDOs are not supported on the HTA hardware, executing
on the AIP runtime defaults to the DSP UDO implementations. HTA
hardware runs exclusively on quantized models and therefore as
with the DSP runtime, a quantized model will be used. The
command to quantize the DLC for AIP is:

cd $SNPE_ROOT/examples/Models/InceptionV3/
    snpe-dlc-quantize --input_dlc dlc/inception_v3_udo.dlc --input_list data/cropped/raw_list.txt --udo_package_path SoftmaxUdoPackage/libs/x86-64_linux_clang/libUdoSoftmaxUdoPackageReg.so --output_dlc dlc/inception_v3_udo_quantized.dlc --enable_hta
    Copy to clipboard

Now push the quantized model to device:

adb push dlc/inception_v3_udo_quantized.dlc /data/local/tmp/inception_v3_udo
    Copy to clipboard

Before executing using the AIP runtime, push the Qualcomm® Neural Processing SDK libraries
for DSP to device with these commands:

adb shell "mkdir -p /data/local/tmp/snpeexample/dsp/lib"
    adb push $SNPE_ROOT/lib/$SNPE_TARGET_DSPARCH/unsigned/*.so /data/local/tmp/snpeexample/dsp/lib
    Copy to clipboard

Now push DSP-specific UDO libraries to device. Depending on DSP
architecture specified in the config, **dsp\_v68** directory can
be **dsp\_v60** or **dsp** (with older Qualcomm® Neural Processing SDK).

cd $SNPE_ROOT/examples/Models/InceptionV3
    adb shell "mkdir -p /data/local/tmp/inception_v3_udo/dsp"
    adb push SoftmaxUdoPackage/libs/dsp_v68/*.so /data/local/tmp/inception_v3_udo/dsp
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libUdoSoftmaxUdoPackageReg.so /data/local/tmp/inception_v3_udo/dsp # Pushes reg lib
    adb push SoftmaxUdoPackage/libs/arm64-v8a/libc++_shared.so /data/local/tmp/inception_v3_udo/dsp
    Copy to clipboard

Then set required environment variables and run snpe-net-run on
device:

adb shell
    cd /data/local/tmp/inception_v3_udo/
    export SNPE_TARGET_ARCH=aarch64-android
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib
    export PATH=$PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin
    export LD_LIBRARY_PATH=/data/local/tmp/inception_v3_udo/dsp/:$LD_LIBRARY_PATH
    export ADSP_LIBRARY_PATH="/data/local/tmp/inception_v3_udo/dsp/;/data/local/tmp/snpeexample/dsp/lib;/system/lib/rfsa/adsp;/system/vendor/lib/rfsa/adsp;/dsp"
    snpe-net-run --container inception_v3_udo_quantized.dlc --input_list target_raw_list.txt --udo_package_path dsp/libUdoSoftmaxUdoPackageReg.so --use_aip
    Copy to clipboard

**Integration with Android APK**

This portion of the tutorial outlines how to integrate Qualcomm® Neural Processing SDK UDO
libraries and Java API for package registration into an Android
application. Generally, for native shared libraries to be
discoverable by the application they must be placed in the
project under

<project>/app/src/main/jniLibs/<platform_abi>
    Copy to clipboard

Once the libraries are accessible by the application, the
registration library can be registered using the provided [Java
API](https://docs.qualcomm.com/doc/80-63442-2/topic/running_model_with_udo.html#executing-neural-networks-with-udo).
This process will be replicated with the example [Image
Classifiers](https://docs.qualcomm.com/doc/80-63442-2/topic/android_tutorial.html#android-sample-application)
application. The following assumes that the rest of the example
application setup has been followed. The tutorial will issue
instructions for platforms with arm64-v8a ABI.

First, create the neccessary directories to contain the UDO
libraries. The following steps will populate all runtime
implementation libraries.

mkdir app/src/main/jniLibs/
    cp -a $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/libs/arm64-v8a/ app/src/main/jniLibs/
    Copy to clipboard

If DSP is to be used as the runtime, copy the implementation
library with the following:

cp $SNPE_ROOT/examples/Models/InceptionV3/SoftmaxUdoPackage/libs/dsp_v68/*.so app/src/main/jniLibs/arm64-v8a/
    Copy to clipboard

If not already done, running **setup\_inceptionv3.sh** will add
the Inception V3 model enabled with UDO and image data to the project.

bash ./setup_inceptionv3.sh
    Copy to clipboard

Now the Java API can be registered.
Include the Qualcomm® Neural Processing SDK AAR which includes necessary native binaries

export SNPE_AAR=snpe-release.aar
    cp $SNPE_ROOT/lib/android/${SNPE_AAR} app/libs/${SNPE_AAR}
    Copy to clipboard

Now the APK can be built and exercised

# exporting gradle user_home is optional but suggested if USER_HOME storage is low
    export GRADLE_USER_HOME=<local path for caching gradle files>
    export ANDROID_SDK_ROOT=<path to android-sdk-9>
    ./gradlew assembleDebug
    Copy to clipboard

For more information, visit [Android Sample App](https://docs.qualcomm.com/doc/80-63442-2/topic/android_tutorial.html#android-sample-application).

Last Published: Oct 02, 2025

[Previous Topic
Build and run a network on ARM64](https://docs.qualcomm.com/bundle/publicresource/80-63442-2/topics/build_samplecode_windows.md) [Next Topic
UDO DSP tutorial for Quantized DLC](https://docs.qualcomm.com/bundle/publicresource/80-63442-2/topics/tutorial_inceptionv3_udo_dsp.md)