Intel® VTune™ Profiler

User Guide

ID 766319
Date 3/22/2024
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents

Configure Yocto Project*/Wind River* Linux* and Intel® VTune™ Profiler with the Intel System Studio Integration Layer

NOTE:
Profiling support for the Yocto Project* is deprecated and will be removed in a future release.

You can use Intel® VTune™ Profiler to collect and analyze performance data on embedded Linux* devices running Yocto Project* or Wind River* Linux*. This example describes how you set up VTune Profiler using the Intel System Studio integration layer, to collect performance data on an embedded device with Yocto Project 1.8 or Wind River* Linux* installed. The integration layer is available with the product installation files. The process integrates the VTune Profiler product drivers with the target package and includes them in the embedded device image. For this example, you need root access to the kernel.

Install the Intel System Studio Integration Layer

Prerequisite: Install Intel System Studio on the host system.

  1. Copy the integration layer from the Intel System Studio installation folder to the appropriate development folder.

    For Yocto Project*:

    cp -r <ISS_BASE_DIR>/YoctoProject/meta-intel-iss <YOCTO_HOME>/

    For Wind River* Linux*:

    cp -r <ISS_BASE_DIR>/YoctoProject/meta-intel-iss <WR_HOME>/
    where
    • <ISS_BASE_DIR> : Root folder of the Intel System Studio installation. By default, this is /opt/intel/system_studio_<version>.x.y/. For example, for the 2019 version, the root folder is /opt/intel/system_studio_2019.0.0/.
    • <YOCTO_HOME> : Root folder of the Yocto Project* cloned directory.
    • <WR_HOME> : Root folder of the Wind River* Linux* cloned directory.
  2. Register the layer by running the post-installation script.

    For Yocto Project*:

    In the shell console, go to the <YOCTO_HOME> folder and run this command:.

    $ meta-intel-iss/yp-setup/postinst_yp_iss.sh <ISS_BASE_DIR>

    For Wind River* Linux*:

    In the shell console, go to the <WR_HOME> folder and run this command:.

    $ meta-intel-iss/yp-setup/postinst_wr_iss.sh <ISS_BASE_DIR>

To uninstall the Intel System Studio integration:

  1. Run the appropriate script to uninstall:

    For Yocto Project*:

    In the shell console, go to the <YOCTO_HOME> folder and run this command:.

    $ meta-intel-iss/yp-setup/uninst_yp_iss.sh 

    For Wind River* Linux*:

    In the shell console, go to the <WR_HOME> folder and run this command:.

    $ meta-intel-iss/yp-setup/uninst_wr_iss.sh 
  2. Remove the meta-intel-iss layer.

Add the Intel VTune Profiler Recipe

  1. Add the path to the wr-iss-<version> to the bblayer.conf file. For example:

    vi /path/to/poky-fido-10.0.0/build/conf/bblayers.conf
    BBLAYERS = "$HOME/source/poky/wr-iss-2019\"

    Your file should look similar to the following:

    BBLAYERS ?= " \
      $HOME/source/poky/meta \
      $HOME/source/poky/meta-poky \
      $HOME/source/poky/meta-yocto-bsp \
      $HOME/source/poky/meta-intel \
      $HOME/source/poky/wr-iss-2019 \
      "
  2. Add the Intel VTune Profiler recipe to conf/local.conf. Two recipes are available,

    intel-vtune-drivers and intel-vtune-sep-driver. In this example, the intel-vtune-drivers is used so the analysis can be run from the VTune Profiler GUI on the host system.

    vi "conf/local.conf"
    IMAGE_INSTALL_append = " intel-vtune-drivers"
    NOTE:

    You cannot add both intel-vtune-drivers and intel-vtune-sep-driver at the same time.

Build and Flash the Target Operating System

  1. Build the target operating system. For example:

    bitbake core-image-sato
  2. Flash the operating system to the embedded device.

Configure and Run Remote Analysis

Use the following steps on the host system to set up and launch the analysis on the embedded device:

  1. Set up a password-less SSH access to the target using RSA keys.

  2. Create a new project.

  3. Select the remote Linux (SSH) analysis system and specify the collection details.

  4. Configure the analysis type.

  5. Start the analysis.