Intel® ESQ for Intel® AI Edge Systems

author-image

By

The Intel® ESQ enables you to test your platform for a specific Intel® Edge Software package using a command line interface tool.

Learn More

Overview

The Intel® Edge System Qualification (Intel® ESQ) is a software package designed to help customers evaluate the performance of AI applications on their specific hardware systems. By running these benchmarks, partners can assess whether their platforms meet the standards required to be qualified as Intel® AI Edge Systems.

For more information, see Edge AI Systems from Intel.

Intel AI Edge Systems Verified Reference Blueprint (VRB)

Verified reference blueprints (VRB) offer optimized configurations for Intel AI Edge Systems, enabling users to fine-tune and benchmark various system types for edge use cases. These blueprints include a comprehensive hardware bill of materials (BOM) and foundational software configurations, such as the operating system, firmware, and drivers. Each blueprint is rigorously tested and validated with a compatible software stack, including frameworks, libraries, orchestration management, and reference use cases. 

Intel ESQ complements this by providing system manufacturers with a reliable method to verify and optimize their solutions, ensuring they meet or exceed VRB guidelines for specific workloads, thereby achieving optimal performance in edge computing environments.

  • Efficiency-optimized edge AI enhances AI performance with integrated CPUs, GPUs, and NPUs, ideal for low power while maximizing potential performance.
  • Scalable performance edge AI delivers adaptable AI performance with a built-in AI accelerator and the option to add a discrete GPU of your choice, including Intel GPUs. 
  • Mainstream and entry-edge AI offer a balance between computing, inferencing, total cost of ownership, and low power.

To learn more, see Intel AI Edge Systems.

 

Recommended Settings for Secure Boot and Performance

Each system configuration may require a specific secure boot and performance configuration. To learn more about BIOS settings, visit Intel AI Edge Systems VRB, and then select a VRB configuration that corresponds to the target platform.

How It Works

This four-step qualification process is designed to be straightforward and user friendly, allowing system manufacturers and integrators to efficiently validate their systems and ensure they are ready for deployment in the market. The qualification not only provides assurance of the system's quality and reliability but also enhances its credibility among users and within the broader ecosystem. Following is an overview. For more details, see the Get Started Guide.

Intel ESQ for Intel AI Edge Systems

Step 1: Download and Install the Qualification Tool

The first step requires registering and downloading the qualification tool. After downloading the app, you must install the qualification tool on the system intended for certification. 

Step 2: Run Qualification Tool 

Once the qualification tool is installed, the next step is to run the program. This involves initiating a series of automated tests that assess various aspects of the system's hardware and software. 

Step 3: Submit Report

Upon completion of the tests, the qualification tool generates a report detailing the system's performance and compatibility results. Submit the report to Intel for approval by emailing intel.edge.ai.systems@intel.com.

Step 4: Publish

Once approved, the newly qualified system is published on the Recommended Systems. 

Get Started

Prerequisites

  1. Install Ubuntu* Desktop 22.04 on the target system.

Note Ubuntu 22.04.3 LTS is preferred for Intel® Core™ and Intel® Core™ Ultra processors. Ubuntu 22.04.4 LTS Server is preferred for Intel® Xeon® processors.

  1. Install required drivers with the corresponding Intel AI Edge Systems configuration as mentioned in the VRB guidelines.
    To speed up system driver and library installation, go to Edge Developer Kit Reference Scripts, and then select the appropriate scripts for the target platform. 
  2. Install libnuma runtime library. 
    sudo apt-get install libnuma-dev python3-dev jq
  3. Set up the Hugging Face* token before running Intel ESQ.
    • Enable Hugging Face. For instructions, refer to the step-by-step guide in the Appendix.
    • Export the Hugging Face token copied in your terminal. Ensure a similar terminal is used to run the application. 
      export HF_TOKEN=<your-huggingface-token>
  4. Set up Docker*. Refer to the official Docker documentation:

Install the Package

  1. To get the qualification tool, go to the Software Catalog page, select your options, and then select Download.
     

    ""

  2. Accept the license agreement and copy the product key.
  3. Transfer the downloaded package to the target Ubuntu system and unzip it: 
    unzip esq_ai_edge_system.zip
  4. Navigate to the extracted package, esq_ai_edge_system, and then change edgesoftware to an executable permission. 
    cd esq_ai_edge_system
chmod +x edgesoftware
  5. Run the following command to install the qualification tool on the target system. 
    ./edgesoftware install
  6. During the installation process, you are prompted to enter the product key. This is the key you copied in step 2, or you can find it in the email from Intel that confirmed your download. 
  7. When the installation is complete, the message Installation of package complete appears with the installation status for each module.
  8. Restart the system.

Run the Intel ESQ Application 

Note Before running Intel ESQ, ensure the qualification tool is downloaded and installed.

 

  1. To run Intel ESQ as a complete test suite using the command line interface (CLI), change the directory to the location of the Intel ESQ folder: 
    cd ~/esq
  2. Check the Intel ESQ version: 
    esq --version
    Ensure the released version matches the version you downloaded. For example: Version: 11.2.0
     
  3. To view the available CLI option, run the following: 
    esq --help
     

  4. To list the available test modules, run the following: 
    esq module list
     

  5. During the test run, to grant permission to your computer, provide the following: 
    • At the start, your Hugging Face account access token

    • Later, your device credentials

Note The system test module is a predistributed test module in the Intel ESQ.

 

  1. Run the qualification test.

Note We recommend running Intel ESQ as a non-root user.

esq --verbose run

 

 

  1. (Optional) To run all the available test modules listed in the package, use the following command: 
    esq run
     

  1. After the test run of the test module is finished, the CLI displays the location of the generated report. Navigate to the Reports folder. 
    cd reports
  2. The report directory displays as shown in the following image:
     

  3. Test results are stored in the output folder. Intel ESQ generates a complete test report in HTML format and detailed logs packaged as one ZIP file, which you can email to the Intel ESQ support team. 

Support

System manufacturers seeking tuning support for Intel AI Edge Systems or inquiries about verified reference blueprints can email intel.edge.ai.systems@intel.com.

Appendix

Enable Intel® Data Center GPU

We recommend that original equipment manufacturers (OEMs) complete the Intel® Data Center GPU Verified System Library program for data center systems designed with Intel Data Center GPUs. This verification process allows customers to validate the functionality of the system for production use. It also provides end-user assurance of system quality before running Intel ESQ tests. 

List of Servers Enabled with Intel® Data Center GPU Flex Series

Enable Hugging Face*

This guide walks you through the process of generating a Hugging Face access token and enabling relevant repository access on Hugging Face.

Generate Hugging Face Access Token

Step 1: Create a Hugging Face Account

  1. Go to Hugging Face.
  2. Sign up or sign in:
    • If you don’t have an account, select Sign Up and follow the instructions to create one.
    • If you already have an account, select Log In and enter your credentials.

Step 2: Generate an Access Token

  1. After you sign in, select your profile picture or username in the top-right corner to open the menu.
  2. From the menu, select Settings.
     

  3. In the Settings menu, select Access Tokens.
     

  4. Select Create new token.
     

  5. Under Token type, select Read and enter a Token name.
     

  6. Select Create token.
  7. Copy and save your access token in a secure location for future use, and then select Done to exit.
  8. Export the Hugging Face token copied in your terminal. Ensure a similar terminal is used to run the application. 
    export HF_TOKEN=<your-huggingface-token>

    Note Remember to copy, export, and store the generated access token so that it can be retrieved in the future.

Enable Relevant Repository Access on Hugging Face

Follow these steps to enable the Llama-3.1-8B-Instruct repository in Hugging Face.

Step 1: Browse Hugging Face Repository

  1. Go to Hugging Face and sign in to your account.
  2. Go to the Llama-3.1-8B-Instruct repository and select Expand to review and access.
     

     

  3. Review the license agreement, fill in the necessary details, accept the license terms, and then select Submit.
     

  4. After submission, validate whether the submission was successful.
     

Step 2: Ensure Successful Access to the Repository

  • When you receive access to the repository, Hugging Face sends you an email. The approval email should be received within one working day. 
     

    Approval email from Hugging Face.

Step 3: Run the Intel AI Edge Systems

After you receive the Llama-3.1-8B-Instruct repository approval email, run the Intel AI Edge Systems application by following the Get Started Guide.

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.