### Navigate to Wheel Package Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/tutorials/AddCustomSample/FrameworkLaunch/CppExtensions/README.md

Change directory to the setup folder for building the custom operator wheel package.

```bash
cd ${git_clone_path}/samples/operator/AddCustomSample/FrameworkLaunch/CppExtensions/setup
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator_contrib/AddCustomSample/KernelLaunch/AddCustomTilingKernel/examples/CPPInvocation/README.md

Before running the example, ensure the operator package has been compiled and deployed. Then, navigate to the directory containing the example code.

```bash
cd ${git_clone_path}/samples/operator_contrib/AddCustomSample/KernelLaunch/AddCustomTilingKernel/examples/CPPInvocation

```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/4_addn_frameworklaunch/AclNNInvocation/README.md

Change the current directory to the AclNNInvocation example directory. This is a prerequisite for running the example.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/4_addn_frameworklaunch/AclNNInvocation
```

--------------------------------

### Navigate to Example Directory (Bash)

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/1_utilities/7_dumptensor/FrameworkLaunch/DumpTensorCube/AclNNInvocation/README.md

Use this command to navigate to the AclNNInvocation example directory within the cloned repository. This is a prerequisite for running the example.

```bash
cd ${git_clone_path}/samples/operator/ascendc/1_utilities/7_dumptensor/FrameworkLaunch/DumpTensorCube/AclNNInvocation
```

--------------------------------

### Install expecttest

Source: https://github.com/ascend/samples/blob/master/operator_contrib/AddCustomSample/KernelLaunch/AddCustomTilingKernel/examples/PythonInvocation/README.md

Install the expecttest library. This is a dependency for the example, likely used for testing or assertion purposes.

```bash
pip3 install expecttest
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/27_simple_add_cpp_extensions/README.md

Change the current directory to the simple add C++ extensions example.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/27_simple_add_cpp_extensions
```

--------------------------------

### Install PyTorch and Torchvision on Atlas200DK

Source: https://github.com/ascend/samples/blob/master/best_practices/contrib/RT-GazeEstimation/Inference/Readme.md

This section details the installation of PyTorch 1.8.1 and torchvision 0.9.1 for the ARM architecture on the Atlas200DK. Refer to the PyTorch installation guide for specific instructions.

```bash
pip3 install pytorch==1.8.1
pip3 install torchvision==0.9.1
```

--------------------------------

### Navigate to the Sample Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/15_sub_frameworklaunch/README.md

Change to the directory containing the install.sh script for the SubCustom operator project.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/15_sub_frameworklaunch
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator_contrib/GreaterEqualSample/KernelLaunch/GreaterEqualKernelInvocation/README.en.md

Change the current directory to the GreaterEqualSample's KernelLaunch/GreaterEqualKernelInvocation directory to begin.

```bash
cd ${git_clone_path}/samples/operator_contrib/GreaterEqualSample/KernelLaunch/GreaterEqualKernelInvocation
```

--------------------------------

### Install torch-npu

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/27_simple_add_cpp_extensions/README.md

Build and install the torch-npu package. This example uses Pytorch 2.1.0, python 3.9, and CANN version 8.0.RC1.alpha002.

```bash
git clone https://gitee.com/ascend/pytorch.git -b v6.0.rc1.alpha002-pytorch2.1.0
cd pytorch/
bash ci/build.sh --python=3.9
pip3 install dist/*.whl
```

--------------------------------

### Navigate to Jit Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/tutorials/AddCustomSample/FrameworkLaunch/CppExtensions/README.md

Change directory to the Jit example folder within the downloaded samples.

```bash
cd ${git_clone_path}/samples/operator/AddCustomSample/FrameworkLaunch/CppExtensions/jit
```

--------------------------------

### Install Dependencies (apt-get)

Source: https://github.com/ascend/samples/blob/master/cplusplus/contrib/TextRecognize/README_CN.md

Installs libeigen3-dev and libjsoncpp-dev using apt-get. Ensure your development and runtime environments have the same architecture or follow specific instructions for cross-architecture setups.

```bash
sudo apt-get install libeigen3-dev libjsoncpp-dev
```

--------------------------------

### Install MMCls and Download ResNet18 Model

Source: https://github.com/ascend/samples/blob/master/inference/contributeSamples/contrib/samplesMMDeploy/README.md

Install the MMCls library and download the ResNet18 model configuration and weights to the 'samples' directory. This is a prerequisite for image classification examples.

```bash
pip3 install mmcls
mim download mmcls --config resnet18_8xb32_in1k --dest samples
```

--------------------------------

### Setup Environment

Source: https://github.com/ascend/samples/blob/master/best_practices/contrib/fsod-code/docs/TRAIN_ON_NOTEBOOK.md

Execute the setup script to configure the necessary environment for training.

```bash
bash setup.sh
```

--------------------------------

### Install FFmpeg for ARM

Source: https://github.com/ascend/samples/blob/master/cplusplus/environment/separate_environmental_guidance_CN.md

Compile and install FFmpeg for an ARM64 environment using a cross-compilation setup. This configuration includes specific flags for cross-compiling and targets the aarch64 architecture.

```bash
./configure --enable-shared --enable-pic --enable-static --disable-x86asm --cross-prefix=aarch64-linux-gnu- --enable-cross-compile --arch=aarch64 --target-os=linux --enable-libx264 --enable-gpl --prefix=${THIRDPART_PATH}
make -j8
make install
```

--------------------------------

### 编译样例

Source: https://github.com/ascend/samples/blob/master/cplusplus/contrib/ar_shadowgan/README_CN.md

执行sample_build.sh脚本进行样例的编译。

```bash
cd $HOME/samples/cplusplus/contrib/ar_shadowgan/scripts
bash sample_build.sh
```

--------------------------------

### Install MMDet and Download Faster R-CNN Model

Source: https://github.com/ascend/samples/blob/master/inference/contributeSamples/contrib/samplesMMDeploy/README.md

Install the MMDet library and download the Faster R-CNN model configuration and weights to the 'samples' directory. This is required for object detection examples.

```bash
pip3 install mmdet
mim download mmdet --config faster_rcnn_r50_fpn_1x_coco --dest samples
```

--------------------------------

### JPEGD Decode Basic Functionality Example

Source: https://github.com/ascend/samples/blob/master/cplusplus/level1_single_api/7_dvpp/jpegd_sample/readme.md

Example command to decode a JPEG image using the JPEGD demo application. It specifies input file, pixel mode, number of channels, starting channel, and send count.

```bash
./jpegd_demo --in_image_file dvpp_jpegd_decode_1920x1080.jpg --pixel_mode 2 --chn_num 2 --start_chn 16 --send_circle 100
```

--------------------------------

### Navigate to Wheel Package Test Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/tutorials/AddCustomSample/FrameworkLaunch/CppExtensions/README.md

Change directory to the test folder within the setup example for running tests.

```bash
cd ${git_clone_path}/samples/operator/AddCustomSample/FrameworkLaunch/CppExtensions/setup/test
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/1_add_frameworklaunch/CppExtensionInvocation/README.md

Change the current directory to the CppExtensionInvocation example directory within the cloned samples repository.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/1_add_frameworklaunch/CppExtensionInvocation
```

--------------------------------

### Run Sample Command

Source: https://github.com/ascend/samples/blob/master/cplusplus/level1_single_api/1_acl/2_memory_management/memcpy_host_device_cpp/README_CN.md

Execute this command to build and run the sample. Adjust parameters like device ID, memory size, and reuse options as needed.

```bash
cd scripts
bash sample_build.sh
cd ../out
./main --release_cycle 10 --number_of_cycles 1 --device_id 0 --memory_size 10485760 --memory_reuse 0 --write_back_host 0
```

--------------------------------

### Run Sample (Save Video Locally)

Source: https://github.com/ascend/samples/blob/master/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera/README_CN.md

Execute the compiled 'main' executable to capture video and save it locally. Ensure the 'output' directory exists before running. Adjust the directory based on your deployment setup.

```bash
# If development and runtime environments are integrated, skip step 1 and proceed directly to step 2.
# 1. Execute the following command to upload the 'ascendcamera' directory from the development environment to the runtime environment (e.g., /home/HwHiAiUser) and log in to the runtime environment (Host) as HwHiAiUser (runtime user).
# "xxx.xxx.xxx.xxx" is the runtime environment IP. For 200DK during USB connection, it is usually 192.168.1.2.
scp -r $HOME/samples/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera HwHiAiUser@xxx.xxx.xxx.xxx:/home/HwHiAiUser
ssh HwHiAiUser@xxx.xxx.xxx.xxx
```

```bash
# 2. Run the executable file.
# If development and runtime environments are integrated, execute the following command to switch directories.
cd $HOME/samples/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera/out
# If development and runtime environments are separated, execute the following command to switch directories.
cd $HOME/ascendcamera/out
# Before running, create the 'output' directory in the 'out' folder.
cd $HOME/ascendcamera/out
mkdir output
# After switching directories, execute the following command to run the sample.
./main
```

--------------------------------

### Install libeigen3-dev on Running Environment (x86 dev, arm run)

Source: https://github.com/ascend/samples/blob/master/cplusplus/contrib/gesturedetection/README_CN.md

For cross-architecture setups (x86 development, ARM execution), install libeigen3-dev on the running environment first. Then, copy the necessary shared objects (so) and header files to the development environment.

```bash
sudo apt-get install libeigen3-dev
```

```bash
# 将arm下的相关的so拷贝到X86的aarch64-linux-gnu目录，不会对本地X86环境本身使用产生任何问题。
cd /usr/lib/aarch64-linux-gnu
# 拷贝相关so，其中X.X.X.X为运行环境ip地址。
sudo scp -r HwHiAiUser@X.X.X.X:/lib/aarch64-linux-gnu/* ./
sudo scp -r HwHiAiUser@X.X.X.X:/usr/lib/aarch64-linux-gnu/* ./
sudo scp -r HwHiAiUser@X.X.X.X:/usr/lib/*.so.* ./
# 拷贝相关头文件
sudo scp -r HwHiAiUser@X.X.X.X:/usr/include/eigen3 /usr/include
```

--------------------------------

### Navigate to Static Library Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/8_library_frameworklaunch/static_library/README.md

Change the current directory to the static library samples directory. This is the starting point for most operations in this guide.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/8_library_frameworklaunch/static_library
```

--------------------------------

### Run Ascendcamera Sample

Source: https://github.com/ascend/samples/blob/master/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera/README_CN.md

Execute the main sample program with specified parameters for video capture, channel, duration, FPS, resolution, and presenter server connection. Replace 'ip:xxxx/presentername' with your actual presenter server IP, port, and a unique view name.

```bash
./main -v -c 1 -t 60 --fps 20 -w 704 -h 576 -s ip:xxxx/presentername
```

--------------------------------

### ResNet50 Quick Start Inference

Source: https://github.com/ascend/samples/blob/master/inference/modelInference/README_CN.md

A foundational example demonstrating the general steps for model inference using AscendCL with the ResNet50 model for object classification.

```cpp
/*
* Copyright (c) 2020, 2023 Huawei Technologies Co., Ltd.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
*     http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/

#include <iostream>
#include <string>
#include <vector>
#include <fstream>

#include "acl/acl.h"
#include "acl/ops/acl_kernel_ops.h"

#include "common/log.h"
#include "common/utils.h"

#include "resnet.h"

#define RESNET50_MODEL_PATH "/home/HwHiAiUser/models/resnet50.om"
#define RESNET50_INPUT_IMAGE_PATH "/home/HwHiAiUser/data/dog1_1024_683.jpg"
#define RESNET50_INPUT_IMAGE_WIDTH 1024
#define RESNET50_INPUT_IMAGE_HEIGHT 683
#define RESNET50_OUTPUT_NUM 5

// ResNet50 model inference
Result ResNet50Infer(const std::string &modelPath, const std::string &imagePath) {
    // Initialize ACL resource
    auto aclRes = std::make_shared<AclResource>();
    Result ret = aclRes->Init();
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to initialize ACL resource.");
        return ret;
    }

    // Load model
    aclmdlDesc *modelDesc = nullptr;
    ret = loadModel(modelPath, &modelDesc);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to load model.");
        return ret;
    }

    // Get model input and output dataset buffer size
    size_t inputSize = 0;
    size_t outputSize = 0;
    ret = getModelInputOutputSize(modelDesc, inputSize, outputSize);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to get model input and output size.");
        return ret;
    }

    // Read image data
    std::vector<char> decodedImage;
    ret = readBinFile(imagePath, decodedImage);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to read image file.");
        return ret;
    }

    // Create input dataset
    aclmdlDataset *inputDataset = aclmdlCreateDataset();
    if (inputDataset == nullptr) {
        ERROR_LOG("Failed to create input dataset.");
        return FAILED;
    }

    // Create input buffer
    void *inputBuffer = nullptr;
    inputBuffer = acldevMalloc(inputSize);
    if (inputBuffer == nullptr) {
        ERROR_LOG("Failed to allocate device buffer for input.");
        return FAILED;
    }

    // Copy image data to device buffer
    ret = acopyHostToDevice(inputBuffer, decodedImage.data(), inputSize);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to copy image data to device.");
        return ret;
    }

    // Add input buffer to dataset
    aclDataBuffer *inputDataBuffer = aclDataBufferMalloc(inputBuffer, inputSize, ACL_MEM_MALLOC_HUGE_FIRST);
    if (inputDataBuffer == nullptr) {
        ERROR_LOG("Failed to create input data buffer.");
        return FAILED;
    }
    ret = aclmdlAddDatasetBuffer(inputDataset, inputDataBuffer);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to add input data buffer to dataset.");
        return ret;
    }

    // Create output dataset
    aclmdlDataset *outputDataset = aclmdlCreateDataset();
    if (outputDataset == nullptr) {
        ERROR_LOG("Failed to create output dataset.");
        return FAILED;
    }

    // Create output buffer
    void *outputBuffer = nullptr;
    outputBuffer = acldevMalloc(outputSize);
    if (outputBuffer == nullptr) {
        ERROR_LOG("Failed to allocate device buffer for output.");
        return FAILED;
    }

    // Add output buffer to dataset
    aclDataBuffer *outputDataBuffer = aclDataBufferMalloc(outputBuffer, outputSize, ACL_MEM_MALLOC_HUGE_FIRST);
    if (outputDataBuffer == nullptr) {
        ERROR_LOG("Failed to create output data buffer.");
        return FAILED;
    }
    ret = aclmdlAddDatasetBuffer(outputDataset, outputDataBuffer);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to add output data buffer to dataset.");
        return ret;
    }

    // Execute model inference
    ret = aclmdlExecute(aclRes->Context(), aclRes->Stream(), modelDesc, inputDataset, outputDataset);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to execute model inference.");
        return ret;
    }

    // Get output data from device
    std::vector<char> inferResult(outputSize);
    ret = acopyDeviceToHost(inferResult.data(), outputBuffer, outputSize);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to copy output data to host.");
        return ret;
    }

    // Process inference result
    ret = processInferResult(inferResult.data(), RESNET50_OUTPUT_NUM);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to process inference result.");
        return ret;
    }

    // Destroy dataset and free buffer
    aclmdlDestroyDataset(inputDataset);
    aclDataBufferFree(inputDataBuffer);
    acldevFree(inputBuffer);
    aclmdlDestroyDataset(outputDataset);
    aclDataBufferFree(outputDataBuffer);
    acldevFree(outputBuffer);

    // Unload model
    ret = unloadModel(modelPath, modelDesc);
    if (ret != SUCCESS) {
        ERROR_LOG("Failed to unload model.");
        return ret;
    }

    INFO_LOG("ResNet50 inference successful.");
    return SUCCESS;
}

int main(int argc, char *argv[]) {
    // Initialize ACL
    aclError ret = aclInit(nullptr);
    if (ret != ACL_ERROR_NONE) {
        ERROR_LOG("Failed to initialize ACL.");
        return -1;
    }

    // Create ACL context
    ret = aclrtCreateContext((void **)&aclRes->Context(), aclRes->DeviceId());
    if (ret != ACL_ERROR_NONE) {
        ERROR_LOG("Failed to create ACL context.");
        aclFinalize();
        return -1;
    }

    // Run ResNet50 inference
    Result inferRet = ResNet50Infer(RESNET50_MODEL_PATH, RESNET50_INPUT_IMAGE_PATH);
    if (inferRet != SUCCESS) {
        ERROR_LOG("ResNet50 inference failed.");
        aclrtDestroyContext(aclRes->Context());
        aclFinalize();
        return -1;
    }

    // Destroy ACL context and finalize ACL
    aclrtDestroyContext(aclRes->Context());
    aclFinalize();

    INFO_LOG("Program execution successful.");
    return 0;
}

```

--------------------------------

### Navigate to Install Script Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/1_add_frameworklaunch/README.md

Change to the directory containing the msOpGen script's install.sh file. This is the starting point for generating and compiling the custom operator.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/1_add_frameworklaunch
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator_contrib/OnesSample/FrameworkLaunch/AclNNInvocation/README.md

Change the current directory to the AclNNInvocation sample directory within the cloned repository.

```bash
cd ${git_clone_path}/samples/operator_contrib/OnesSample/FrameworkLaunch/AclNNInvocation
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/0_introduction/21_vectoradd_kernellaunch/VectorAddMultiCoreWithTiling/README.md

Use this command to navigate to the VectorAddMultiCoreWithTiling example directory. Replace ${git_clone_path} with the actual path where the samples repository was cloned.

```bash
cd ${git_clone_path}/samples/operator/ascendc/0_introduction/21_vectoradd_kernellaunch/VectorAddMultiCoreWithTiling/
```

--------------------------------

### Download Calibration Data

Source: https://github.com/ascend/samples/blob/master/python/level1_single_api/9_amct/amct_tensorflow/cmd/README_CN.md

Use this command to download sample calibration data and labels for quantization. Ensure you are in the example's root directory and have the 'unrar' tool installed.

```bash
cd data
mkdir image && cd image
wget https://obs-9be7.obs.cn-east-2.myhuaweicloud.com/models/amct_acl/classification/calibration.rar
unrar e calibration.rar
```

--------------------------------

### Navigate to Sample Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/4_best_practices/15_mata_address_conflict/AclNNInvocation/README.md

Use this bash command to navigate to the AclNNInvocation sample directory for running the example.

```bash
cd ${git_clone_path}/samples/operator/ascendc/4_best_practices/15_mata_address_conflict/AclNNInvocation
```

--------------------------------

### Download and Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/4_best_practices/22_matmul_reduce_scatter_custom/AclNNInvocation/README.md

This bash command downloads the sample code and navigates into the AclNNInvocation directory for the MatmulReduceScatterCustom operator. Ensure you have git installed and are in the desired parent directory.

```bash
cd ${git_clone_path}/samples/operator/ascendc/4_best_practices/22_matmul_reduce_scatter_custom/AclNNInvocation
```

--------------------------------

### Compile the Sample

Source: https://github.com/ascend/samples/blob/master/inference/ACLHelloWorld/README.md

Navigate to the sample's script directory and execute the build script to compile the sample.

```bash
cd $HOME/samples/inference/ACLHelloWorld/scripts
bash sample_build.sh
```

--------------------------------

### Configure .bashrc for Integrated Setup

Source: https://github.com/ascend/samples/blob/master/docs/INSTALL_cn.md

Set environment variables for CPU architecture, third-party library paths, and CANN installation directory. This is for users running development and execution on the same Ascend device.

```bash
# 以安装用户在任意目录下执行以下命令，打开.bashrc文件。
vi ~/.bashrc  
# 在文件最后一行后面添加如下内容。
export CPU_ARCH=`arch`
export THIRDPART_PATH=${HOME}/Ascend/thirdpart/${CPU_ARCH}  #代码编译时链接samples所依赖的相关库文件
export LD_LIBRARY_PATH=${THIRDPART_PATH}/lib:$LD_LIBRARY_PATH  #运行时链接第三方库文件
export INSTALL_DIR=${HOME}/Ascend/ascend-toolkit/latest #CANN软件安装后的文件存储路径，根据安装目录自行修改
# 执行命令保存文件并退出。
:wq!  
# 执行命令使其立即生效。 
source ~/.bashrc 
# 创建samples第三方相关依赖文件夹
mkdir -p ${THIRDPART_PATH}
# 下载源码并安装git
cd ${HOME}
sudo apt-get install git
git clone https://gitee.com/ascend/samples.git
# 拷贝公共文件到samples相关依赖路径中
cp -r ${HOME}/samples/inference/acllite/aclliteCPP ${THIRDPART_PATH}
```

--------------------------------

### Build Sample

Source: https://github.com/ascend/samples/blob/master/cplusplus/level2_simple_inference/0_data_process/cropandpaste/README_CN.md

Execute the build script to compile the sample. Ensure you are in the correct directory.

```bash
cd ${HOME}/samples/cplusplus/level2_simple_inference/0_data_process/cropandpaste/scripts    
bash sample_build.sh
```

--------------------------------

### Source CANN Environment Setup Script

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/1_utilities/0_printf/FrameworkLaunch/README.md

Source the setenv.bash script to export necessary CANN environment variables, including ASCEND_OPP_PATH, if it is not already set. Replace [ASCEND_INSTALL_PATH] with the actual CANN installation path.

```bash
source [ASCEND_INSTALL_PATH]/bin/setenv.bash
# 例如 source /usr/local/Ascend/ascend-toolkit/latest/bin/setenv.bash
```

--------------------------------

### Navigate to Example Directory

Source: https://github.com/ascend/samples/blob/master/operator_contrib/FastGeluGradSample/FrameworkLaunch/AclNNInvocation/README.en.md

Change the current directory to the AclNNInvocation example directory.

```bash
cd ${git_clone_path}/samples/operator_contrib/FastGeluGradSample/FrameworkLaunch/AclNNInvocation
```

--------------------------------

### Configure .bashrc for Separated Setup (Runtime Environment)

Source: https://github.com/ascend/samples/blob/master/docs/INSTALL_cn.md

Set environment variables for CPU architecture, third-party library paths, and CANN installation directory on the runtime Ascend device. Ensure CPU_ARCH matches the device's architecture.

```bash
# 以安装用户在运行环境任意目录下执行以下命令，打开.bashrc文件。
vi ~/.bashrc  
# 在文件最后一行后面添加如下内容。CPU_ARCH环境变量请根据运行环境cpu架构填写，如export CPU_ARCH=aarch64
export CPU_ARCH=`arch`
export THIRDPART_PATH=${HOME}/Ascend/thirdpart/${CPU_ARCH}  #代码编译时链接第三方库
export LD_LIBRARY_PATH=${HOME}/Ascend/thirdpart/${CPU_ARCH}/lib:$LD_LIBRARY_PATH  #运行时链接库文件
export INSTALL_DIR=${HOME}/Ascend/ascend-toolkit/latest  #CANN软件安装后文件存储路径
# 执行命令保存文件并退出。
:wq!  
# 执行命令使其立即生效。 
source ~/.bashrc 
# 创建第三方依赖文件夹
mkdir -p ${THIRDPART_PATH}
# 拷贝相关数据，其中X.X.X.X为开发环境ip地址。
sudo scp -r HwHiAiUser@X.X.X.X:${THIRDPART_PATH}/common ${THIRDPART_PATH}
```

--------------------------------

### Navigate to Load Library Example Directory

Source: https://github.com/ascend/samples/blob/master/operator/ascendc/tutorials/AddCustomSample/FrameworkLaunch/CppExtensions/README.md

Change directory to the load_library example folder for compiling custom C++ APIs.

```bash
cd ${git_clone_path}/samples/operator/AddCustomSample/FrameworkLaunch/CppExtensions/load_library
```

--------------------------------

### Configure .bashrc for Separated Setup (Development Environment)

Source: https://github.com/ascend/samples/blob/master/docs/INSTALL_cn.md

Set environment variables for CPU architecture, third-party library paths, and CANN installation directory on the development machine. Adjust CPU_ARCH and THIRDPART_PATH based on the target runtime environment.

```bash
# 以安装用户在开发环境任意目录下执行以下命令，打开.bashrc文件。
vi ~/.bashrc  
# 在文件最后一行后面添加如下内容。CPU_ARCH环境变量请根据运行环境cpu架构填写，如export CPU_ARCH=aarch64
export CPU_ARCH=[aarch64/x86_64]
# THIRDPART_PATH需要按照运行环境安装路径设置，如运行环境为arm，指定安装路径为Ascend-arm，则需要设置为export THIRDPART_PATH=${HOME}/Ascend-arm/thirdpart/${CPU_ARCH}
export THIRDPART_PATH=${HOME}/Ascend/thirdpart/${CPU_ARCH}  #代码编译时链接第三方库
# CANN软件安装后文件存储路径，最后一级目录请根据运行环境设置，运行环境为arm，这里填arm64-linux；运行环境为x86，则这里填x86_64-linux，以下以arm环境为例
export INSTALL_DIR=${HOME}/Ascend/ascend-toolkit/latest/arm64-linux
# 执行命令保存文件并退出。
:wq!  
# 执行命令使其立即生效。 
source ~/.bashrc 
# 创建第三方依赖文件夹
mkdir -p ${THIRDPART_PATH}
# 拷贝公共文件到第三方依赖文件夹
cd $HOME
git clone https://gitee.com/ascend/samples.git
cp -r ${HOME}/samples/inference/acllite/aclliteCPP ${THIRDPART_PATH}
```

--------------------------------

### Run Sample (Save Image Locally)

Source: https://github.com/ascend/samples/blob/master/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera/README_CN.md

Execute the compiled 'main' executable with specific parameters to capture an image and save it locally. Adjust the directory based on your deployment setup.

```bash
# If development and runtime environments are integrated, skip step 1 and proceed directly to step 2.
# 1. Execute the following command to upload the 'ascendcamera' directory from the development environment to the runtime environment (e.g., /home/HwHiAiUser) and log in to the runtime environment (Host) as HwHiAiUser (runtime user).
# "xxx.xxx.xxx.xxx" is the runtime environment IP. For 200DK during USB connection, it is usually 192.168.1.2.
scp -r $HOME/samples/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera HwHiAiUser@xxx.xxx.xxx.xxx:/home/HwHiAiUser
ssh HwHiAiUser@xxx.xxx.xxx.xxx
```

```bash
# 2. Run the executable file.
# If development and runtime environments are integrated, execute the following command to switch directories.
cd $HOME/samples/cplusplus/level1_single_api/5_200dk_peripheral/ascendcamera/out
# If development and runtime environments are separated, execute the following command to switch directories.
cd $HOME/ascendcamera/out
# After switching directories, execute the following command to run the sample.
./main -i -c 1 -o ./output/filename.jpg --overwrite
```

--------------------------------

### Execute PNGD Demo (Basic Functionality)

Source: https://github.com/ascend/samples/blob/master/cplusplus/level1_single_api/7_dvpp/pngd_sample/readme.md

Run the pngd_demo executable to decode a PNG image. This example decodes the specified image into RGB_888 format using 2 channels starting from channel 16, and performs the decoding 100 times.

```bash
./pngd_demo --in_image_file dvpp_pngd_1920x1080_RGBA.png  --img_width 1920  --img_height 1080 --pixel_mode 12 --chn_num 2 --start_chn 16 --send_circle 100 
```

--------------------------------

### Build Sample

Source: https://github.com/ascend/samples/blob/master/cplusplus/level2_simple_inference/2_object_detection/VGG_SSD_coco_detection_CV_with_AIPP/README_CN.md

Execute this script to compile the sample. Ensure you are in the 'scripts' directory within the sample's folder.

```bash
cd ${HOME}/samples/cplusplus/level2_simple_inference/2_object_detection/VGG_SSD_coco_detection_CV_with_AIPP/scripts    
bash sample_build.sh
```