Intel® oneAPI Toolkits Installation

Guide for Linux* OS

Intel® oneAPI Toolkits Installation Guide for Linux* OS

Contents
Chapter 1: Intel® oneAPI Toolkits and Components Installation
Guide for Linux* OS
Prerequisites.............................................................................................3
Install Intel GPU Drivers..................................................................... 4
GPU: Disable Hangcheck .................................................................... 4
Installation ...............................................................................................5
Install with GUI.................................................................................6
Install with Command Line ................................................................. 6
Install Using Package Managers ........................................................ 11
YUM/DNF/Zypper .................................................................... 12
APT....................................................................................... 15
Conda ................................................................................... 17
PIP ....................................................................................... 21
NuGet ................................................................................... 23
Cloudera................................................................................ 25
Maven ................................................................................... 25
Spack.................................................................................... 25
List Available Toolkits, Components, and Runtime Library
Packages ........................................................................... 26
Install Packages or Components to Different Directories ....................... 27
Configure WSL 2 for GPU Workflows .................................................. 28
Install Software for Intel FPGA Development Flows .............................. 30
Install the Intel® Quartus Prime Software ................................... 32
Install Intel® FPGA Board Packages............................................ 33
Use oneAPI Components in a Yocto Project Build ................................. 33
Uninstall oneAPI Toolkits and Components .................................................. 34
Troubleshooting....................................................................................... 36
Notices and Disclaimers............................................................................ 38

2
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1

Intel® oneAPI Toolkits and

Components Installation Guide
for Linux* OS 1
This guide covers the installation of Intel® oneAPI toolkits and standalone components on Linux* OS.

Before You Begin

Check the Prerequisites information before installing the Intel oneAPI packages.

Installation
Install Intel oneAPI component and toolkit packages with one of the following options:
• Install with GUI
• Install with Command Line
• Install Using Package Managers
• Install Using Docker Container

Next Steps
Get Intel oneAPI code samples and refer to the toolkit Get Started page for detailed usage instructions,
examples, and more:
• Get Started with Intel® oneAPI Base Toolkit
• Get Started with Intel® oneAPI HPC Toolkit
• Get Started with Intel® oneAPI IoT Toolkit
• Get Started with Intel® oneAPI AI Analytics Toolkit
• Get Started with Intel® oneAPI System Bring-up Toolkit
• Get Started with Intel® oneAPI Rendering Toolkit
• Get Started with Intel® oneAPI DL Framework Developers Toolkit

Prerequisites
Consider the following important information before installing the Intel oneAPI packages.

System Requirements
Refer to the toolkit-specific Release Notes and System Requirements documents to learn more about
compatibility details:
• Intel® oneAPI Base Toolkit Release Notes | System Requirements
• Intel® oneAPI HPC Toolkit Release Notes | System Requirements
• Intel® oneAPI IoT Toolkit Release Notes | System Requirements
• Intel® oneAPI AI Analytics Toolkit Release Notes | System Requirements
• Intel® oneAPI System Bring-up Toolkit Release Notes | System Requirements
• Intel® oneAPI Rendering Toolkit Release Notes | System Requirements
• Intel® oneAPI DL Framework Developers Toolkit Release Notes | System Requirements

3
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

IDE Integration
To use third-party IDE, install Eclipse* on your Linux* OS host system before installing oneAPI Toolkits. This
allows you to integrate the plugins as part of the Intel oneAPI Base Toolkit installation.

Use Modulefiles or a Configuration File to Set Environment Variables (optional)

Environment variables can be set up manually (as described in Get Started Guides and sample README files)
or automatically using one of the methods below: - Use modulefiles - Use a setvars.sh configuration file

Set Up Your System for Intel GPU

If you are using Intel GPU, complete the following steps before or after Intel oneAPI installation:
• Install Intel GPU drivers
• Check that you have fulfilled the requirements of Intel® Graphics Compute Runtime for oneAPI Level Zero
and OpenCL™ Driver. Make sure that you have permissions to access the /dev/dri/renderD\*
and /dev/dri/card\* files. This typically means that your user account is a member of the video (on
Ubuntu* 18, Fedora* 30, and SLES* 15 SP1) or render (on Ubuntu* 19 and higher, CentOS* 8, and
Fedora* 31) group. Alternatively, an administrator with sudo or root privilege can change the group owner
of /dev/dri/renderD\* and /dev/dri/card\* to a group ID used by your user base.
• If you have applications with long-running GPU compute workloads in native environments, you must
disable the hangcheck timeout period to avoid terminating workloads.
• If you plan to use the Intel® Distribution for GDB* on Linux* OS, make sure to configure debugger access.
• Install Intel GPU Drivers
• GPU: Disable Hangcheck

Install Intel GPU Drivers

If you use Intel GPU, you need to install the latest GPU drivers separately. To install the driver packages,
follow the Installation Guide applicable for your Linux distribution.

GPU: Disable Hangcheck

This section applies only to applications with long-running GPU compute workloads in native environments. It
is not recommended for virtualizations or other standard usages of GPU, such as gaming.
A workload that takes more than four seconds for GPU hardware to execute is a long-running workload. By
default, individual threads that qualify as long-running workloads are considered hung and are terminated.
By disabling the hangcheck timeout period, you can avoid this problem.

NOTE If the system is rebooted, hangcheck is automatically enabled. You must disable hangcheck
again after every reboot or follow the directions below to disable hangcheck persistently (across
multiple reboots). Please re-run this GPU Hangcheck disable with reboot fix if you update (or auto-
update) the kernel.

To disable hangcheck until the next reboot:

sudo sh -c "echo N> /sys/module/i915/parameters/enable_hangcheck"
To disable hangcheck across multiple reboots:

4
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
NOTE If the kernel is updated, hangcheck is automatically enabled. Run the procedure below after
every kernel update to ensure hangcheck is disabled.

1. Open a terminal.
2. Open the grub file in /etc/default.
3. In the grub file, find the line GRUB_CMDLINE_LINUX_DEFAULT="".
4. Enter the following text between the quotes (“”):
i915.enable_hangcheck=0
5. Run the following command:
sudo update-grub
6. Reboot the system. Hangcheck remains disabled.

Installation

Toolkit Installation

Important Some domain-specific toolkits require you to install the Intel oneAPI Base Toolkit first for
full functionality.

Use one of the following options to install a toolkit package:

Toolkit Name Binary Installer Package Manager Docker Container

Intel® oneAPI Base Online/offline installer YUM, DNF, Zypper, APT Docker Hub
Toolkit

Intel® oneAPI HPC Online/offline installer YUM, DNF, Zypper, APT Docker Hub
Toolkit (requires
installation of the Intel
oneAPI Base Toolkit)

Intel® oneAPI IoT Toolkit Online/offline installer YUM, DNF, Zypper, APT Docker Hub
(requires installation of
the Intel oneAPI Base
Toolkit)

Intel® oneAPI AI Online/offline installer YUM, DNF, Zypper, APT, Docker Hub
Analytics Toolkit Conda

Intel® oneAPI Rendering Online/offline installer YUM, DNF, Zypper, APT N/A
Toolkit

Each of the binary installer types operates in various modes, which are covered later in this section.

NOTE When using the offline installer, you can enable full offline mode by setting the environment
variable INTEL_SUPPRESS_INTERNET_CONNECTION=1. When this mode is enabled:

• Installer does not send installation statistics

• Download-only mode is disabled
• Upgrade mode is disabled

5
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Component Installation
You can install toolkit components as standalone via binary installer or package managers. Refer to the
Single Component Downloads and Runtime Versions resource to locate a component package.

Intel® FPGA Developmental Flow

For emulation and FPGA optimization report flow, you can just use the Intel oneAPI Base Toolkit that includes
the Intel oneAPI DPC++/C++ Compiler. For more information, see FPGA Flow in the Intel oneAPI
Programming Guide.
For the Intel FPGA hardware or simulation flow, you can target Intel® FPGA devices, Intel® Programmable
Acceleration Cards (PACs), or a third-party vendor-provided BSP.
• To target Intel® FPGA devices or custom platforms, see Install Software for Intel FPGA Development
Flows.
• To target third party vendor-provided BSP and the FPGA Platform, see Intel® FPGA Add-on for oneAPI
Base Toolkit. The vendor specifies which version of Intel Quartus Prime software is necessary to compile
the BSP. You must install the Intel® Quartus Prime software and BSP packages separately.

Install with GUI

After downloading the toolkit installation package, follow the steps below to install it with GUI.
1. Launch the installer with the following command:
• root: sudo sh ./l_[Toolkit Name]Kit_[version].sh
• user: sh ./l_[Toolkit Name]Kit_[version].sh
2. Follow the installer instructions.
3. Once the installation is complete, verify that your toolkit is installed to the correct installation directory:
• root: /opt/intel/oneapi
• user: ~/intel/oneapi

NOTE If you are using Intel GPU, you need to install the latest GPU drivers separately.

Install with Command Line

Command line installation supports the following installation modes:

• Interactive mode prompts you to select or confirm certain options during the installation process
• Non-interactive (silent) mode allows you to define the installation configuration only once and does not
require any user input during installation
The general instructions below are common for both modes.

General Instructions
Launch the installation script using the following command:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh [options] -a [arguments]

• user:

sh ./l_[Toolkit Name]Kit_[version].sh [options] -a [arguments]

6
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
where [options] contain parameters for the package extraction script, and [arguments] are options for
the installer.
The package extraction script supports the following options:

-h, --help Show help for the package extraction script.

-f, --extract-folder Point to the folder where the package content will
be saved.

-x, --extract-only This option unpacks the installation package only. It

does not launch the installer.

-r, --remove-extracted-files <yes|no> Remove extracted files after installation. This action
cleans up the temporary package file location.

-l, --log <log file> Log all package extraction actions to the specified
file.

-a <arguments> Pass arguments to the installer.

The values after -a are passed as command line arguments to the installer. The following installer options
are supported:

Option Supported Default Description

mode value (if
option is not
passed)

-c, --cli CLI N/A Run the installer in interactive text-based user
interface (TUI) mode.

-s, --silent Silent N/A Run the installer in non-interactive (silent) mode.

--eula Silent decline Required. Accept or decline End User License

Agreement (EULA), supported values: accept or
decline (default).

--action Silent/CLI install Specify one of the supported values below when
the installer action is needed:
• install (default) Install the product. Use the
--components option to specify the list of
components to be installed. If not specified, the
default set of components is installed.
• remove Uninstall the product.
• modify Change the current set of components
installed. List all the components you need using
the --components option. Components that are
already installed still must be in the list if remain
relevant.
• downloadonly Download an offline installation
package without installing it. To customize the
list of components to be included into a
package, use the --components option.
• repair Repair the currently installed product.

7
1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Option Supported Default Description

mode value (if
option is not
passed)

--instance Silent/CLI default Specify an ID of an installation instance. For

example: sh ./l_[Toolkit
Name]Kit_[version].sh -a --
instance=<instance ID>. This option enables
side-by-side installation of oneAPI products. Each
instance is a separate installation entity with its
own isolated environment. Product installed in one
instance is not visible in another instance. If
omitted, installation is performed in default
instance. To get the list of available instances, use
the --list-instances option.

--list- Silent/CLI N/a Get the list of available installation instances.

instances

--config Silent/CLI N/A Point to the configuration INI file with options. You
can use this file as an alternative to passing options
via the command line; mixed approach is also
supported. Sample content of a configuration file:
s=eula=accept.
Use this command to run the installer with the
options passed via config.txt:

sh ./l_[Toolkit Name]Kit_[version].sh --
config config.txt

--components Silent default Specify components to perform an action on,

supported values: all, default, custom
components split by ‘:’. If you need the default
components and some extra component(s),
combine default with the name of the extra
component(s) separated by ‘:’. For example: --
components default:<component_name>.

--list-products N/A N/A Get the list of downloaded products, their IDs,
versions and statuses (installed/not installed). Use
together with the --instance option to get the list
of available products in a specific instance. For
example: sh ./l_[Toolkit
Name]Kit_[version].sh -a --list-products
--instance=<instance ID>.

--product-id Silent/CLI N/A Specify an ID of a product to perform an action on.

Use this option with --list-components or --
action {install|remove|modify|repair}.

--product-ver Silent/CLI N/A Specify a product version to perform an action on.

Use this option with --list-components or --
action {install|remove|modify|repair}.

8
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Option Supported Default Description
mode value (if
option is not
passed)

--list- N/A N/A Get the list of available components of the current
components package or of a product specified with --product-
id. Use together with the --instance option to
get the list of available components in a specific
instance. For example: sh ./l_[Toolkit
Name]Kit_[version].sh -a --list-
components --instance=<instance ID>.

--package-path Silent/CLI N/A Specify the directory of the package to install.

--install-dir Silent default Supported in silent mode. Customize the

installation installation directory.
directory

--log-dir Silent/CLI default log Customize the directory to save the log file to.
location

--proxy Silent/CLI N/A Specify proxy settings in the following format:

http://username:password@proxy-
server.mycorp.com:3128.

--download- Silent default Point to the directory to store all downloaded and
cache download cached files.
cache
location

--download-dir Silent default Customize the download directory, which is used in

download download-only mode.
directory

--intel-sw- Silent decline Accept or decline participation in Intel Software

improvement- Improvement Program, supported values: accept
program-consent or decline (default). To get the program
description, use the --show-intel-sw-
improvement-program-consent command.

--show-intel- N/A N/A Show the detailed description of the Intel Software
sw-improvement- Improvement Program.
program-consent

--ignore-errors Silent/CLI N/A Complete installation even if non-critical errors

occur. Check the log file for the list of errors
occured and ignored during installation.

-h, --help N/A N/A Show the installer help.

-p, --property Silent/CLI N/A Pass additional custom options. For example, the
string -p=option1=value -p option2=value
gives two additional options. If a custom option is
provided twice with different values, only the latest

9
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Option Supported Default Description

mode value (if
option is not
passed)

one will be used. For example, the string -

p=option=a -p option=b takes b as value for
option.

For example, to show the installer help, use the following command:

sh ./l_[Toolkit Name]Kit_[version].sh -a -h

Non-interactive (Silent) Installation

1. Use the following command to launch the installer in silent mode:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh -a --silent --eula accept

• user:

sh ./l_[Toolkit Name]Kit_[version].sh -a --silent --eula accept

For the full list of supported command line options, refer to the General Instructions section.
2. Once the installation is complete, verify that the toolkit is installed in the default directory:
• root:

/opt/intel/oneapi
• user:

~/intel/oneapi

NOTE If you are using Intel GPU, you need to install the latest GPU drivers separately.

Interactive Installation
1. Launch the installer with the following command:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh -a --cli

• user:

sh ./l_[Toolkit Name]Kit_[version].sh -a --cli

For the full list of supported command line options, refer to the General Instructions section.
1. Follow the installer instructions.
2. Once the installation is complete, verify that your toolkit is installed to the correct directory:
• root:

/opt/intel/oneapi
• user:

~/intel/oneapi

10
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Examples
• Display the list of already installed products and products included in the downloaded package:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh -a --silent --list-products

• user:

sh ./l_[Toolkit Name]Kit_[version].sh -a --list-products

Example of output:

ID Version Language Installed Name

============================================================================================
intel.oneapi.lin.tbb.product 2021.1.1-129 false Intel® oneAPI Threading Building Blocks
• Display the list of components in product of current package:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh -a --list-components

• user:

sh ./l_[Toolkit Name]Kit_[version].sh -a --list-components

• Display the list of components of any installed product on the system:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh -a --list-components --product-id

intel.oneapi.lin.tbb.product --product-ver 2021.1.1-129
• user:

sh ./l_[Toolkit Name]Kit_[version].sh -a --list-components --product-id

intel.oneapi.lin.tbb.product --product-ver 2021.1.1-129
Example of output:

ID Version Language Installed Name

============================================================================================
intel.oneapi.lin.tbb.devel 2021.1.1-129 Intel® oneAPI Threading Building Blocks
• Install specific Intel oneAPI Toolkit products and components:
• root:

sudo sh ./l_[Toolkit Name]Kit_[version].sh --silent --eula accept --components

intel.oneapi.lin.tbb.devel
• user:

sh ./l_[Toolkit Name]Kit_[version].sh -a --silent --eula accept --components

intel.oneapi.lin.tbb.devel

Install Using Package Managers

You can install Intel oneAPI packages from one of the following repositories:
• YUM, DNF, Zypper
• APT
• Cloudera (oneMKL)
• Spack
• Conda
• PIP
• NuGet
• Maven (oneDAL)

11
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

For instructions on how to get the list of packages available for installation with Linux* package managers,
refer to List Toolkits, Components, and Runtime Library Packages.

YUM/DNF/Zypper

You can install Intel oneAPI packages via YUM, DNF, or Zypper package managers (whichever works best for
your system configuration).

Pre-installation Steps

NOTE
• it with the following command:
If you have an existing
installation of Intel® # If using YUM or DNF:
oneAPI Beta, remove sudo -E {yum|dnf} autoremove <package_name>

# If using Zypper:
sudo -E zypper rm <package_name>
• When upgrading from 2021.1 to 2021.2, apply automatic removal of conflicting packages during
the upgrade process as described in the Upgrade Toolkit/Component section.

1. Check the toolkit-specific System Requirements page to make sure that your OS is supported:
• Intel® oneAPI Base Toolkit
• Intel® oneAPI HPC Toolkit
• Intel® oneAPI IoT Toolkit
• Intel® oneAPI AI Analytics Toolkit
• Intel® oneAPI Rendering Toolkit
• Intel® oneAPI DL Framework Developers Toolkit
You can get your OS version using the following command depending on your Linux distribution:

# Redhat, Fedora, CentOS and related

more /etc/redhat-release

# Ubuntu, Debian, others

more /etc/lsb-release
2. If you plan to use Intel GPU, install the Intel GPU drivers.
3. Set up the repository:
If using YUM/DNF:
a. Create the YUM or DNF repo file in the /temp directory as a normal user:

tee > /tmp/oneAPI.repo << EOF

[oneAPI]
name=Intel® oneAPI repository
baseurl=https://yum.repos.intel.com/oneapi
enabled=1
gpgcheck=1
repo_gpgcheck=1
gpgkey=https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
EOF

12
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
b. Move the newly created oneAPI.repo file to the YUM/DNF configuration directory /etc/
yum.repos.d:

sudo mv /tmp/oneAPI.repo /etc/yum.repos.d

If using Zypper:
Add the Intel oneAPI repository public key with the following command:

sudo zypper addrepo https://yum.repos.intel.com/oneapi oneAPI

By adding this new repository, Zypper had to automatically import the public repo key. For some cases
rpm might require explicit key import by:

rpm --import https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB

Install Packages

NOTE If you are on a company intranet or behind a firewall, set the http_proxy and https_proxy
environment variables to allow YUM/DNF/Zypper access the repository servers using HTTPS protocol.

1. Get the name of a toolkit package that you need to install from the list of Intel oneAPI packages. Write
down or copy your package name for future reference.
2. Install the needed package with the following command:

sudo {yum|dnf|zypper} install <package_name>

For example, to install the Intel® oneAPI Base Toolkit package from YUM, use:

sudo yum install intel-basekit

If you need to install on a machine with no internet access, or in case of a large distributed installation
on a cluster, you can download a package without installing it with the --download-only option. For
more details, refer to YUM instructions or Zypper man pages.
Installation is complete! For next steps, refer to the Get Started Guide for your toolkit:
• Intel® oneAPI Base Toolkit
• Intel® oneAPI HPC Toolkit
• Intel® oneAPI IoT Toolkit
• Intel® oneAPI AI Analytics Toolkit
• Intel® oneAPI Rendering Toolkit
• Intel® oneAPI DL Framework Developers Toolkit

NOTE If you have applications with long-running GPU compute workloads in native environments, you
must disable the hangcheck timeout period to avoid terminating workloads.

13
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Upgrade Toolkit/Component

NOTE Intel oneAPI packages of version 2021.2 (and higher) have conflicts with 2021.1 packages.
Before installing 2021.2, you need to remove the 2021.1 packages from your system. If you use YUM
v4/DNF/Zypper, you can apply automatic removal of conflicting packages during upgrade with the
following commands:

# If using YUM:
sudo yum upgrade --allowerasing --best <package_name>

# If using DNF:
sudo dnf upgrade --allowerasing --best <package_name>

# If using Zypper:
sudo zypper update --force-resolution <package_name>
For lower versions of YUM, remove the 2021.1 packages manually:

sudo yum autoremove <package_name>

For more information about the issue and workarounds, refer to the YUM/DNF and ZYPPER Packages
oneAPI 2021.1 Gold (Initial Release) issue will Prevent Upgrades article.
To re-install 2021.1 packages, use web and local installers as described in the Installation section of
this document.

You can upgrade toolkit or component package to the latest version using the following instructions:
• Toolkit: {yum|dnf|zypper} upgrade <toolkit package>

For example, to upgrade the Intel oneAPI Base Toolkit package to the latest version, use the following
command:

# If using YUM:
sudo yum upgrade intel-basekit

# If using DNF:
sudo dnf upgrade intel-basekit

# If using Zypper:
sudo zypper upgrade intel-basekit
• Component: {yum|dnf|zypper} upgrade <component package>

For example, to upgrade the Intel Distribution for GDB* package, use the following command:

# If using YUM:
sudo yum upgrade intel-oneapi-dpcpp-debugger

# If using DNF:
sudo dnf upgrade intel-oneapi-dpcpp-debugger

# If using Zypper:
sudo zypper upgrade intel-oneapi-dpcpp-debugger

List of Intel® oneAPI Packages

14
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Toolkit Packages
The following toolkits and associated versions are available for installation via YUM repositories:

NOTE The repositories always contain the latest released version.

Toolkit Name 64-bit Meta Package Name 32-bit Meta Package Name*
(default)

Intel® oneAPI Base Toolkit intel-basekit intel-basekit-32bit

Intel® oneAPI HPC Toolkit intel-hpckit intel-hpckit-32bit

Intel® oneAPI IoT Toolkit intel-iotkit intel-iotkit-32bit

Intel® oneAPI DL Framework intel-dlfdkit intel-dlfdkit-32bit

Developer Toolkit

Intel® AI Analytics Toolkit intel-aikit intel-aikit-32bit

Intel® oneAPI Rendering Toolkit intel-renderkit intel-renderkit-32bit

* - only required if you deploy and deploy 32-bit applications

Intel® System Bring-Up Toolkit is not distributed via a repository, see details.
Intel® Distribution of OpenVINO™ toolkit for Linux* is distributed via separate YUM and APT repositories.

Runtime Library Packages

The oneAPI repository provides runtime library packages. Install these packages on systems where you run
oneAPI applications but do not do development, compilation, or runtime profiling. The following runtime
library packages are available:
• oneAPI runtime libraries package, which is a superset of all runtimes for oneAPI components:
• 64-bit: intel-oneapi-runtime-libs
• 32-bit: intel-oneapi-runtime-libs-32bit
• Component runtime library packages. For instructions on how to get the list of all available standalone
runtime packages, refer to the List Standalone Runtime Library Packages section.

APT

Pre-installation Steps

NOTE If you have an existing installation of Intel® oneAPI Beta, remove it with the following
command:

sudo apt autoremove <package_name>

1. Check the toolkit-specific System Requirements page to make sure that your OS is supported:
• Intel® oneAPI Base Toolkit
• Intel® oneAPI HPC Toolkit
• Intel® oneAPI IoT Toolkit

15
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

• Intel® oneAPI AI Analytics Toolkit

• Intel® oneAPI Rendering Toolkit
• Intel® oneAPI DL Framework Developers Toolkit
You can get your OS version using the following command depending on your Linux distribution:

# Redhat, Fedora, CentOS and related

more /etc/redhat-release

# Ubuntu, Debian, others

more /etc/lsb-release
2. If you plan to use Intel GPU, install the Intel GPU drivers.
3. Set up the repository:

# download the key to system keyring

wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB \
| gpg --dearmor | sudo tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null

# add signed entry to apt sources and configure the APT client to use Intel repository:
echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/
oneapi all main" | sudo tee /etc/apt/sources.list.d/oneAPI.list
4. Update packages list and repository index:

sudo apt update

Install Packages

NOTE If you are on a company intranet or behind a firewall, set the http_proxy and https_proxy
environment variables to allow APT access the repository servers using HTTPS protocol.

1. Get the name of a toolkit package that you need to install from the list of Intel oneAPI packages. Write
down or copy your package name for future reference.
2. Install the needed package with the following command:

sudo apt install <package_name>

For example, to install the Intel® oneAPI Base Toolkit package, use:

sudo apt install intel-basekit

#repeat 'apt install ...' for each toolkit you need
If you need to install on a machine with no internet access, or in case of a large distributed installation
on a cluster, you can download a package without installing it with the --download-only option.

NOTE If you want to integrate tools into the Eclipse* IDE, open Eclipse and verify that a menu titled
Intel is present. If the menu is not present, see Installing Eclipse* Plugins from the IDE.

Installation is complete! For next steps, refer to the Get Started Guide for your toolkit:
• Intel® oneAPI Base Toolkit
• Intel® oneAPI HPC Toolkit
• Intel® oneAPI IoT Toolkit
• Intel® oneAPI AI Analytics Toolkit
• Intel® oneAPI Rendering Toolkit
• Intel® oneAPI DL Framework Developers Toolkit

16
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
NOTE If you have applications with long-running GPU compute workloads in native environments, you
must disable the hangcheck timeout period to avoid terminating workloads.

List of Intel® oneAPI Packages

Toolkit Packages
The following toolkits and associated versions are available for installation via APT repositories:

NOTE The repositories always contain the latest released version.

Toolkit Name 64-bit Meta Package Name 32-bit Meta Package Name*
(default)

Intel® oneAPI Base Toolkit intel-basekit intel-basekit-32bit

Intel® oneAPI HPC Toolkit intel-hpckit intel-hpckit-32bit

Intel® oneAPI IoT Toolkit intel-iotkit intel-iotkit-32bit

Intel® oneAPI DL Framework intel-dlfdkit intel-dlfdkit-32bit

Developer Toolkit

Intel® AI Analytics Toolkit intel-aikit intel-aikit-32bit

Intel® oneAPI Rendering Toolkit intel-renderkit intel-renderkit-32bit

* - only required if you deploy and deploy 32-bit applications

Intel® System Bring-Up Toolkit is not distributed via a repository, see details.
Intel® Distribution of OpenVINO™ toolkit for Linux* is distributed via separate YUM and APT repositories.

Runtime Library Packages

The oneAPI repository provides runtime library packages. Install these packages on systems where you run
oneAPI applications but do not do development, compilation, or runtime profiling. The following runtime
library packages are available:
• oneAPI runtime libraries package, which is a superset of all runtimes for oneAPI components:
• 64-bit: intel-oneapi-runtime-libs
• 32-bit: intel-oneapi-runtime-libs-32bit
• Component runtime library packages. For instructions on how to get the list of all available standalone
runtime packages, refer to the List Standalone Runtime Library Packages section.

Conda

This page provides general instructions on installing the Intel® oneAPI component packages via the Conda*
package manager.
For additional installation notes, refer to the Conda documentation.
To install a package, execute the following command:
• To install the latest version available:
conda install -c intel <package_name>

17
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

To get your package name, refer to the list of packages in the table below.
• To install a specific version:
conda install -c intel <package_name>==<version>
For example: conda install -c intel mkl==2021.1.1

List of Available Packages

Component Name Package Name Platfo Dependencies

rm

Intel® MPI Library impi_rt linux N/A

-x64
impi-devel

Intel® Fortran Compiler (Beta) and Intel® intel-fortran-rt linux Intel® MPI Library
Fortran Compiler Classic -x64
Intel OpenMP* Runtime
linux Library
-x86

Intel® CPU Runtime for OpenCL™ intel-opencl-rt linux oneTBB

Applications -x64

Intel® oneAPI DPC++/C++ Compiler dpcpp-cpp-rt linux Intel® CPU Runtime for
-x64 OpenCL™ Applications
Intel OpenMP* Runtime
Library

Intel® oneAPI DPC++ Library onedpl-devel linux N/A

-x64

Intel OpenMP* Runtime Library intel-openmp linux N/A

-x64
linux
-x86

Intel® oneAPI Threading Building Blocks tbb linux N/A

(oneTBB) -x64
tbb-devel
linux
-x86

tbb4py linux N/A

-x64

Intel® oneAPI Data Analytics Library dal linux oneTBB

(oneDAL) -x64
dal-static
linux
dal-devel
-x86
dal-include

daal4py linux oneDAL

-x64

18
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Component Name Package Name Platfo Dependencies
rm

Intel® Extension for Scikit-learn scikit-learn- linux oneDAL

intelex -x64

Intel® Integrated Performance Primitives ipp linux N/A

(Intel® IPP) -x64
ipp-static
linux
ipp-include
-x86

Intel® Integrated Performance Primitives ipp_crypto linux N/A

Cryptography -x64
ipp_crypto-static
ipp_crypto-
include

Intel® oneAPI Math Kernel Library mkl linux Intel OpenMP* Runtime
(oneMKL) -x64 Library
mkl-devel
linux oneTBB
mkl-static
-x86
mkl-include

Intel® oneAPI Deep Neural Network onednn-cpu-gomp linux N/A

Library (oneDNN)* -x64
onednn-devel-cpu-
gomp

onednn-cpu-iomp linux Intel OpenMP* Runtime

-x64 Library
onednn-devel-cpu-
iomp

onednn-cpu-dpcpp- linux Intel® CPU Runtime for

gpu-dpcpp -x64 OpenCL™ Applications

onednn-devel-cpu- Intel oneAPI DPC++/C++

dpcpp-gpu-dpcpp Compiler Runtime

Intel® oneAPI Collective Communications oneccl-devel linux N/A

Library (oneCCL) -x64

* - For Intel® oneAPI Deep Neural Network Library (oneDNN), only packages of identical configuration can be
installed into one environment. For example, you can install onednn-devel-cpu-vcomp with onednn-cpu-
vcomp, but should avoid installing it with packages of other configurations, like cpu-iomp, cpu-tbb, cpu-
dpcpp-gpu-dpcpp.
• Install Intel® AI Analytics Toolkit via Conda*
• List of Available Packages

Install Intel® AI Analytics Toolkit via Conda*

Intel provides access to the AI Kit through a public Anaconda repository. If you do not have an existing
Conda-based python environment, install Conda and Miniconda*. To get more details on the AI Analytics
Toolkit, visit the Intel AI Analytics toolkit home page.

19
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

The AI Kit contains three distinct Python environments targeting different use cases:
• intel-aikit-tensorflow for deep learning workflows using Intel® Optimization for TensorFlow*.
• intel-aikit-pytorch for deep learning workflows using Intel® Optimization for PyTorch*.
• intel-aikit-modin for data analytics and machine learning workflows using Intel® Distribution of Modin
(for accelerated Panda data frames), Intel® Extension for Scikit-learn* and Intel optimizations for XGboost
(for ML training and inference).
• intel-aikit for data science workstation development. The oneAPI AI kit samples and documents are
not applicable to the intel-aikit Conda package.

NOTE To get the latest version of the Intel(R) Optimization for TensorFlow*, you must first install
Python 3.9, then install the AI Kit through Anaconda.

To instal the AI Kit via Conda, complete the following steps:

1. Activate your existing python conda environment located in <pythonhome>:

source <pythonhome>/bin/activate
2. Install the AI Kit oneAPI packages in a new environment using conda create. A list of available
packages is located at https://anaconda.org/intel/repo. Not all packages in the Anaconda repository are
up to date with the current release. If the repo contains an outdated version of a required component,
get a newer one by installing via the command line or GUI.
If the repository contains the desired version, create an AI Kit Tensorflow* environment named aikit-
tf with this version:
conda create -n aikit-tf -c intel intel-aikit-tensorflow
Similarly, you can create an AI Kit PyTorch environment named aikit-pt:

conda create -n aikit-pt -c intel intel-aikit-pytorch

You can also create an AI Kit Modin and machine learning environment named aikit-modin:
For Python 3.7:
conda create -n aikit-modin -c intel intel-aikit-modin
For other Python versions, refer to the instructions in Intel® Distribution of Modin Getting Started Guide.
3. Set user environment. After the toolkit is installed, before accessing the tools, you must activate your
python environment and set up environment variables to access the tools. For example, to activate the
python environment created in the previous step, use:
conda activate aikit-tf

NOTE
• To install the Model Zoo for Intel® Architecture component of the toolkit, clone the main branch to
your local directory: git clone https://github.com/IntelAI/models.git.
• If you have applications with long-running GPU compute workloads in native environments, you
must disable the hangcheck timeout period to avoid terminating workloads.

List of Available Packages

NOTE Intel® packages are available on intel label on the Anaconda* Cloud. You must include -c
intel on your command line as in the examples above, or add intel to your Conda configuration file
using conda config --add channels intel.

20
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Component Name Package Name Platform

Intel® Distribution for Python* intelpython3_full linux-x64

Intel® Distribution of Modin* (via Anaconda intel-aikit-modin linux-x64

distribution of the toolkit using the Conda package
manager)

Intel® Neural Compressor neural-compressor linux-x64

Intel® Optimization for PyTorch* intel-aikit-pytorch linux-x64

Intel® Optimization for TensorFlow* intel-aikit- linux-x64

tensorflow

After you have installed your components, view the Get Started Guide for the Intel oneAPI AI Analytics
Toolkit to build and run a sample or explore Getting Started Samples on GitHub.

PIP

This page provides general instructions on installing the Intel® oneAPI component packages from the Python*
Package Index (PyPI).
For additional installation notes, refer to the PyPI documentation.
To install a package, execute the following command:
• To install the latest version available:
pip install <package_name>
To get your package name, refer to the list of packages in the table below.
• To install a specific version:
pip install -c intel <package_name>==<version>
For example: pip install mkl==2021.1.1

Important For Intel® oneAPI Deep Neural Network Library (oneDNN), only packages of identical
configuration can be installed into one environment. For example, you can install onednn-devel-cpu-
vcomp with onednn-cpu-vcomp, but should avoid installing it with packages of other configurations,
like cpu-iomp, cpu-tbb, cpu-dpcpp-gpu-dpcpp.

List of Available Packages

Component Name Package Name Platfo Dependencies

rm

Intel® MPI Library impi_rt linux N/A

-x64
impi-devel

Intel® Fortran Compiler (Beta) and Intel® intel-fortran-rt linux Intel® MPI Library
Fortran Compiler Classic -x64
Intel OpenMP* Runtime
linux Library
-x86

21
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Component Name Package Name Platfo Dependencies

rm

Intel® CPU Runtime for OpenCL™ intel-opencl-rt linux oneTBB

Applications -x64

Intel® oneAPI DPC++/C++ Compiler dpcpp-cpp-rt linux Intel® CPU Runtime for
-x64 OpenCL™ Applications
Intel OpenMP* Runtime
Library

Intel OpenMP* Runtime Library intel-openmp linux N/A

-x64
linux
-x86

Intel® oneAPI Threading Building Blocks tbb linux N/A

(oneTBB) -x64
tbb-devel
linux
-x86

tbb4py linux N/A

-x64

Intel® oneAPI Data Analytics Library daal linux oneTBB

(oneDAL) -x64
daal-static
linux
daal-devel
-x86
daal-include

daal4py linux oneDAL

-x64

Intel® Extension for Scikit-learn scikit-learn- linux oneDAL

intelex -x64

Intel® Integrated Performance Primitives ipp linux N/A

(Intel® IPP) -x64
ipp-static
linux
ipp-include
-x86

Intel® Integrated Performance Primitives ipp_crypto linux N/A

Cryptography -x64
ipp_crypto-static
linux
ipp_crypto-
-x86
include

Intel® oneAPI Math Kernel Library mkl linux Intel OpenMP* Runtime
(oneMKL) -x64 Library
mkl-devel
linux oneTBB
mkl-static
-x86
mkl-include

22
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Component Name Package Name Platfo Dependencies
rm

mkl-dpcpp linux Intel OpenMP* Runtime

-x64 Library
mkl-devel-dpcpp
linux oneTBB
-x86 Intel® oneAPI DPC++/C++
Compiler Runtime
Intel® CPU Runtime for
OpenCL™ Applications

Intel® oneAPI Deep Neural Network onednn-cpu-gomp linux N/A

Library (oneDNN) -x64
onednn-devel-cpu-
gomp

onednn-cpu-iomp linux Intel OpenMP* Runtime

-x64 Library
onednn-devel-cpu-
iomp

onednn-cpu-dpcpp- linux Intel® CPU Runtime for

gpu-dpcpp -x64 OpenCL™ Applications

onednn-devel-cpu- Intel oneAPI DPC++/C++

dpcpp-gpu-dpcpp Compiler Runtime

Intel® oneAPI Collective Communications oneccl-devel linux N/A

Library (oneCCL) -x64

Intel® Distribution of Modin* (via intel-aikit-modin linux N/A

Anaconda distribution of the toolkit using -x64
the Conda package manager)

Intel® Neural Compressor neural-compressor linux N/A

-x64

Intel® Optimization for TensorFlow* intel-tensorflow linux N/A

-x64
intel-tensorflow-
avx512

NuGet

This page provides general notes on how to install Intel® oneAPI components distributed via the NuGet
channel. NuGet is a Microsoft-supported mechanism for sharing compiled code. It also defines how the
packages are created, hosted and consumed, and it provides the tools for each of those roles. For more
details on the installation process, please refer to the Microsoft* documentation.
Intel® oneAPI components distributed via NuGet include both development and runtime options.
For your convenience, the components are divided to devel and static packages corresponding to the
different linking types (dynamic and static). Certain component packages are also split into x64 and x86
versions to reduce the overall package size.

Development Packages
The following table provides the full list of available packages:

23
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Component Name Package Name Platfo Dependencies

rm

Intel® MPI Library intelmpi.devel.<pl linux N/A

atform> -x64

Intel OpenMP* Runtime Library intelopenmp.devel. linux N/A

<platform>
intelopenmp.static
.<platform>

Intel® oneAPI Data Analytics Library inteldal.devel.<pl linux N/A

(oneDAL) atform> -x64
inteldal.static.<p
latform>

Intel® oneAPI Threading Building Blocks inteltbb.devel.<pl linux N/A

(oneTBB) atform>

Intel® Integrated Performance Primitives intelipp.devel.<pl linux N/A

atform> -x64
intelipp.static.<p linux
latform> -x86
intelipp.nonpic.<p
latform>

Intel® Integrated Performance Primitives intelipp_crypto.de linux N/A

Cryptography vel.<platform> -x64
intelipp_crypto.st linux
atic.<platform> -x86
intelipp_crypto.no
npic.<platform>

Intel® oneAPI Math Kernel Library intelmkl.devel.<pl linux N/A

atform> -x64
intelmkl.static.<p linux
latform> -x86

Intel® oneAPI Math Kernel Library intelmkl.devel.clu linux Intel OpenMP* Runtime
(Cluster Components) ster.<platform> -x64 Library

intelmkl.static.cl oneMKL
uster.<platform>

All the specified dependencies will be downloaded automatically by the NuGet Package Manager.

Runtime Packages
The runtime packages are runtime redistributable libraries that will automatically load optimizations specific
to your Intel hardware (including, but not limited to, vectorization). They can be used by another NuGet
package that depends on these runtimes.

24
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Component Name Package Name Platform
Availability

Intel® oneAPI Video Processing Library (oneVPL) onevpl.runtime.<platfor linux-x64

m>

Cloudera

This page provides general instructions on installing the Intel® oneAPI component parcel packages using the
Cloudera* Manager. For additional notes, refer to the Parcels documentation and Cloudera Installation Guide.
Currently, only the Intel® oneAPI Math Kernel Library (oneMKL) component is distributed via Cloudera*.
Package name: mkl-<version>-el7.parcel, for example, mkl-2021.2.0.296-el7.parcel.
The install the oneMKL parcel:
1. In the Cloudera Manager Admin Console, access the Parcels page by doing one of the following:
• Click the Parcels indicator in the left navigation bar.
• Click the Hosts in the left navigation bar, then click the Parcels tab.
2. At the Parcels page, click the Parcel Repositories & Network Settings button.
3. In the Remote Parcel Repository URLs list, click the plus symbol to open an additional row. Enter
the path to Intel® MKL Parcel repository: http://parcels.repos.intel.com/mkl/latest. Click the
Save & Verify configuration button.
4. Click the Check for New Parcels button. In the Location selector, click Available Remotely. The
latest oneMKL parcel should be available for download.
5. Click the Download button for the oneMKL parcel. By downloading the package, you agree with the
terms and conditions stated in the End-User License Agreement (EULA).
6. When download is completed, click the Distribute button to distribute the parcel on all cluster nodes.
7. When distribution is completed, click the Activate button to activate the parcel on all cluster nodes.
Note
The repository URL referenced above installs the latest version of oneMKL parcel. To install a lower version,
use the URL based on the following model: http://parcels.repos.intel.com/mkl/
<version>.<update>.<build_number>.

Maven

This page provides general instructions on how to include Intel® oneAPI Data Analytics Library (oneDAL)
packages from the Maven repository into your Java project.
To enable oneDAL in your project, specify the following artifacts in your build automation tool:
Group ID: com.intel.dal
Version: <version>
Artifact ID: dal
where <version> is a valid component version, for example, 2021.2.0.123.
For more information on the Maven dependency mechanism, refer to the Maven documentation.

Spack

This page provides general instructions on installing the Intel® oneAPI component packages via Spack. After
installation, you can use the tools directly or use Spack to build packages with the tools.
To install a package, execute the following command:
spack install <package_name>

25
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Example
The example below demonstrates how to set up Intel oneAPI compilers with Spack. For the full list of oneAPI
packages available via Spack, refer to the Package List.
1. Install the compilers package with the following command:

spack install intel-oneapi-compilers

2. Add the oneAPI compilers to the set of compilers that Spack can use:

spack compiler add `spack location -i intel-oneapi-compilers`/compiler/latest/linux/bin/intel64

spack compiler add `spack location -i intel-oneapi-compilers`/compiler/latest/linux/bin
This adds the compilers to your compilers.yaml.
3. Verify that the compilers are available:

spack compiler add

The intel-oneapi-compilers package includes two families of compilers:

• Intel: icc, iccpc, ifort - Intel’s classic compilers

• oneAPI: icx, icpx, ifx - Intel’s new generation of compilers based on LLVM

To build the patchelf Spack package with icc, use:

spack install patchelf%intel

To build with icx, use:

spack install patchelf%oneapi

In addition to compilers, oneAPI contains many libraries. The hdf5 package works with any compatible MPI
implementation. To build hdf5 with Intel oneAPI MPI, use:

spack install hdf5 +mpi

^intel-oneapi-mpi
For more information, see Spack documentation.

List Available Toolkits, Components, and Runtime Library Packages

Use the commands provided below to find and install specific toolkits, standalone components, standalone
runtime library packages, or simply to see all available packages in a corresponding oneAPI repository:
• List toolkit packages
• List standalone components
• List standalone runtime library packages
• List all packages

List Toolkit Packages

To query the repository for available toolkit packages, use the following command:
YUM/DNF:

sudo -E {yum|dnf} --disablerepo="*" --enablerepo="oneAPI" list available | grep kit | grep -v

runtime
Zypper:

sudo -E zypper pa -ir oneAPI | grep kit | grep -v runtime

APT:

sudo -E apt-cache pkgnames intel | grep kit | grep -v runtime

26
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1

List Standalone Components

The oneAPI repository also contains standalone components, which are packages that provide a specific tool
for cases where you do not need an entire toolkit. For these packages, if there is a <component>-runtime
package, make sure to get and install both the component package and its runtime package. Not all
standalone components need an additional runtime package. If you do not see a runtime package for your
standalone component, then you do not need one.
To query the repository for available standalone components and their runtime packages, use the following
command:
YUM/DNF:

sudo -E {yum|dnf} --disablerepo="*" --enablerepo="oneAPI" list available | grep intel-oneapi |

grep -v intel-oneapi-runtime
Zypper:

sudo -E zypper pa -ir oneAPI | grep intel-oneapi | grep -v intel-oneapi-runtime

APT:

sudo -E apt-cache pkgnames intel | grep intel-oneapi | grep -v intel-oneapi-runtime

List Standalone Runtime Library Packages

The oneAPI repository provides standalone runtime library packages. Install these packages on systems
where you run oneAPI applications but do not do development, compilation, or runtime profiling. In this case,
you only need the shared libraries dynamically linked to by executables, provided by these packages.
To query the repository for available component runtime libraries, use the following command:
YUM/DNF:

sudo -E {yum|dnf} --disablerepo="*" --enablerepo="oneAPI" list available | grep intel-oneapi-

runtime
Zypper:

sudo -E zypper pa -ir oneAPI | grep intel-oneapi-runtime

APT:

sudo -E apt-cache pkgnames intel | grep intel-oneapi-runtime

List All Packages

To query all available Intel® oneAPI packages provided in a repository, use the following command:
YUM/DNF:

sudo -E {yum|dnf} --disablerepo="*" --enablerepo="oneAPI" list available

Zypper:

sudo -E zypper pa -ir oneAPI

APT:

sudo -E apt-cache pkgnames intel

Install Packages or Components to Different Directories

27
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Intel oneAPI installer supports side-by-side installation. It means that you can install multiple instances of
toolkits/components to different directories on the same machine. Each instance is a separate installation
entity with its own isolated environment. Product installed in one instance is not visible in another instance.
With multi-instance installation, you can:
• Install a newer version of a toolkit/component without removing the previous one
• Install toolkits to different directories other than default
• Have multiple instances of the same version of a toolkit/component installed
To install a package into a specific instance, use the following command:
• root:
sudo sh ./l_[Toolkit Name]Kit_[version].sh -a -s –eula=accept –-install-dir=<custom-install-dir>
--instance=<instance ID>
• user:
./l_[Toolkit Name]Kit_[version].sh -a -s –eula=accept –-install-dir=<custom-install-dir> --
instance=<instance ID>
where
• <custom-install-dir> is the directory where you want to install a specific instance to
• <instance ID> is a unique combination of alphanumeric symbols that designate an instance, for
example my-custom-instance-1
For instructions on how to uninstall product(s) from a specific instance, refer to Uninstall Using Silent CLI.

Configure WSL 2 for GPU Workflows

With Microsoft* Windows Subsystem for Linux 2 (WSL 2), you can use native Linux distribution of Intel®
oneAPI tools and libraries on Windows*. Get the latest version of WSL 2 following the process described in
The Windows Subsystem for Linux in the Microsoft Store is now generally available on Windows 10 and 11.
To be able to use Intel oneAPI tools on WSL 2 for GPU workflows, install the Intel GPU drivers as described
below.

Ubuntu* 20.04 (focal)

Step 1: Add package repository
Install the repositories.intel.com/graphics package repository by executing the following command in
your WSL 2 console:

sudo apt-get install -y gpg-agent wget

wget -qO - https://repositories.intel.com/graphics/intel-graphics.key |
sudo gpg --dearmor --output /usr/share/keyrings/intel-graphics.gpg
echo 'deb [arch=amd64 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://
repositories.intel.com/graphics/ubuntu focal-devel main' | \
sudo tee /etc/apt/sources.list.d/intel.gpu.focal.list
sudo apt update

Tip Before pasting the command to your console, run sudo ls and enter your password to prevent
the commands from being swallowed by the sudo password prompt.

The code above performs the following:

• Checks that your system has gpg-agent and wget installed
• Downloads and installs the public key used to verify the integrity of the package repository

28
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
• Adds the repositories.intel.com/graphics repository to the system

Step 2: Install runtime and development (optional) packages

Install GPU software packages with the following command:

sudo apt-get install \

intel-opencl-icd \
intel-level-zero-gpu level-zero \
intel-media-va-driver-non-free libmfx1 libmfxgen1 libvpl2 \
libegl-mesa0 libegl1-mesa libegl1-mesa-dev libgbm1 libgl1-mesa-dev libgl1-mesa-dri \
libglapi-mesa libgles2-mesa-dev libglx-mesa0 libigdgmm11 libxatracker2 mesa-va-drivers \
mesa-vdpau-drivers mesa-vulkan-drivers va-driver-all
OPTIONAL: If you plan to perform development tasks, you need to install the following optional development
packages to make sure that oneAPI tools function correctly:

sudo apt-get install -y \

libigc-dev \
intel-igc-cm \
libigdfcl-dev \
libigfxcmrt-dev \
level-zero-dev
Reboot WSL 2 by executing wsl.exe –shutdown in PowerShell.

Step 3: Verify installation

Verify Computing drivers installation:

sudo apt-get install clinfo

clinfo
The command should return similar to the following:

Number of platforms 2
Platform Name Intel(R) OpenCL HD Graphics
Platform Vendor Intel(R) Corporation
Platform Version OpenCL 3.0
Platform Profile FULL_PROFILE

Ubuntu* 22.04 (jammy)

Step 1: Add package repository
Install the repositories.intel.com/graphics package repository by executing the following command in
your WSL 2 console:

sudo apt-get install -y gpg-agent wget

wget -qO - https://repositories.intel.com/graphics/intel-graphics.key |
sudo gpg --dearmor --output /usr/share/keyrings/intel-graphics.gpg
echo 'deb [arch=amd64,i386 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://
repositories.intel.com/graphics/ubuntu jammy arc' | \
sudo tee /etc/apt/sources.list.d/intel.gpu.jammy.list

Tip Before pasting the command to your console, run sudo ls and enter your password to prevent
the commands from being swallowed by the sudo password prompt.

The code above performs the following:

• Checks that your system has gpg-agent and wget installed

29
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

• Downloads and installs the public key used to verify the integrity of the package repository
• Adds the repositories.intel.com/graphics repository to the system

Step 2: Install runtime and development (optional) packages

Install GPU software packages with the following command:

sudo apt-get install -y \

intel-opencl-icd intel-level-zero-gpu level-zero \
intel-media-va-driver-non-free libmfx1 libmfxgen1 libvpl2 \
libegl-mesa0 libegl1-mesa libegl1-mesa-dev libgbm1 libgl1-mesa-dev libgl1-mesa-dri \
libglapi-mesa libgles2-mesa-dev libglx-mesa0 libigdgmm12 libxatracker2 mesa-va-drivers \
mesa-vdpau-drivers mesa-vulkan-drivers va-driver-all
OPTIONAL: If you plan to perform development tasks, you need to install the following optional development
packages to make sure that oneAPI tools function correctly:

sudo apt-get install -y \

libigc-dev \
intel-igc-cm \
libigdfcl-dev \
libigfxcmrt-dev \
level-zero-dev
To support Steam games, install i386 packages:

sudo dpkg --add-architecture i386

sudo apt-get update

sudo apt-get install -y \

udev mesa-va-drivers:i386 mesa-common-dev:i386 mesa-vulkan-drivers:i386 \
libd3dadapter9-mesa-dev:i386 libegl1-mesa:i386 libegl1-mesa-dev:i386 \
libgbm-dev:i386 libgl1-mesa-glx:i386 libgl1-mesa-dev:i386 \
libgles2-mesa:i386 libgles2-mesa-dev:i386 libosmesa6:i386 \
libosmesa6-dev:i386 libwayland-egl1-mesa:i386 libxatracker2:i386 \
libxatracker-dev:i386 mesa-vdpau-drivers:i386 libva-x11-2:i386
Reboot WSL 2 by executing wsl.exe –shutdown in PowerShell.

Step 3: Verify installation

Verify Computing drivers installation:

sudo apt-get install clinfo

clinfo
The command should return similar to the following:

Number of platforms 2
Platform Name Intel(R) OpenCL HD Graphics
Platform Vendor Intel(R) Corporation
Platform Version OpenCL 3.0
Platform Profile FULL_PROFILE

Install Software for Intel FPGA Development Flows

Field-programmable gate arrays (FPGAs) are configurable integrated circuits that you can program to
implement arbitrary circuit topologies. Classified as spatial compute architectures, FPGAs differ significantly
from fixed Instruction Set Architecture (ISA) devices such as CPUs and GPUs. FPGAs offer a different set of

30
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
optimization trade-offs from these traditional accelerator devices. While you can compile SYCL* code for CPU,
GPU or FPGA, the compiling process for FPGA development is somewhat different than that for CPU or GPU
development.
SYCL supports accelerators in general. The Intel® oneAPI DPC++/C++ Compiler implements additional FPGA-
specific support to assist FPGA code development. For additional information about the FPGA flows, refer to
Types of SYCL* FPGA Compilation topic in the Intel oneAPI Programming Guide.

How to Work With FPGA?

The following sections describe various methods you can work with FPGA:

Use Preinstalled Environment

If you are new to oneAPI and FPGA development, then Intel recommends using the Intel® DevCloud. Intel®
DevCloud provides a preinstalled development environment with free access to Intel® oneAPI toolkits and
components, latest Intel® hardware, optimized frameworks, tools, and libraries to speed up your learning and
project prototyping.
Intel® DevCloud is already set up with an Intel® Programmable Acceleration Card (PAC) with Intel Arria® 10
GX FPGA and Intel® FPGA PAC D5005 (previously known as Intel® PAC for Intel® Stratix® 10 SX FPGA) and the
necessary software stack. For more information, refer to Get Started with Intel® oneAPI Base Toolkit on the
DevCloud.

Set Up Your Own System and Install Software

If you want to set up your own system, then use one of these methods:
• Set up a single system: In this method, you can use a single system acting as both the runtime and
development system. Install the Intel oneAPI Base Toolkit, Intel® Quartus® Prime Software, and custom
platform/FPGA device (hardware-run, machine-specific) on the same system.
• Set up separate development and runtime systems: In this method, you install the custom platform/
FPGA device on the runtime system and run only the design. On the development system, install the
Intel® Quartus® Prime software to compile and generate the FPGA bitstream. Refer to Intel® Quartus®
Prime Software, Install Intel® FPGA Board Packages sections for more information.
• Set up a cloud on-premise: A cloud on-premise helps reduce the hardware cost necessary for
development. In this workflow, you can set up two development systems, one for the FPGA development
and the other for the Intel® Quartus® Prime software compilation. The runtime system can be different.
After setting up your development systems, install the physical card on the system. Refer to the Intel®
oneAPI DPC++ /C++ Library System Requirements for FPGA requirements.
• On the first development system with lower configurations (8 GB RAM), iterate over your designs using
the emulation and report flow to verify code correctness. You just need to install the Intel® oneAPI
Base Toolkit package on this system. For more information about emulation and report flow, refer to
the Types of FPGA Compilation.
• On the second development system with higher configurations (48 or 64 GB RAM based on the Intel
PAC you use), install the Intel® Quartus® Prime Software. Perform Intel® Quartus® Prime compilation
using either the hardware flow or the device link flow. For more information, refer to the FPGA Flow in
the Intel® oneAPI Programming Guide.

31
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

NOTE
• FPGA IP Authoring flow is now supported. Developing IP components/hardware with oneAPI
requires the Intel oneAPI Base Toolkit and Intel® Quartus® Prime Software. For details about getting
started with the IP component development flow, refer to Getting Started with Intel® oneAPI
Toolkits and Intel® Quartus® Prime Software and FPGA Flow in the Intel® oneAPI Programming
Guide.
• Intel® Quartus® Prime Software is required only for simulation and hardware generation flows, and
integrating your IP component into your design. You can generate reports and RTL code, and run
the emulation stage with only the Intel® oneAPI Base Toolkit.

For additional details, refer to the following topics:

• Install the Intel® Quartus Prime Software
• Install Intel® FPGA Board Packages

Install the Intel® Quartus Prime Software

The Intel® Quartus® Prime software includes everything you need to design for Intel® FPGAs, from design
entry and synthesis to optimization, verification, and simulation. It contains the following features:
• Hybrid Placer & Global Router
• Timing Analyzer
• Physical Synthesis
• Incremental Fitter Optimization
• Interface Planner
• Synthesis Engine
• Platform Designer
• Partial Reconfiguration
• Block-Based (Hierarchical) Design

NOTE The Intel® Quartus® Prime software is not required for the FPGA development flow’s emulation
or report generation stages. You can complete those stages with just the Intel® oneAPI DPC++/C++
Compiler included in the Intel oneAPI Base Toolkit.

If you want to use the Intel® Quartus® Prime software (required for FPGA hardware and simulation flow) with
oneAPI, the edition of Intel® Quartus® Prime software that you need depends on your target device (either
standalone device or the device on an acceleration board). Refer to the Intel® oneAPI DPC++ /C++ Library
System Requirements for more information.
You have the following options for using Intel® Quartus® Prime software with oneAPI:
• If you want to install a version of the Intel® Quartus® Prime software, follow the instructions in the
following documents:
• Intel® FPGA Software Installation and Licensing
• Intel Quartus Prime Pro Edition User Guide: Getting Started
• Intel Quartus Prime Standard Edition User Guide: Getting Started
• If you already have a version of the Intel® Quartus® Prime software installed on your system and you want
to use that version, then use one of the following methods to set up the environment:
• Set QUARTUS_ROOTDIR_OVERRIDE = <path_to_your_quartus_folder>
• Add the bin directory of the Intel® Quartus® Prime software to your PATH variable.
• If you have multiple versions of the Intel® Quartus® Prime software installed, Intel recommends setting
the QUARTUS_ROOTDIR_OVERRIDE variable to point to the Quartus software path you want to use.
Otherwise, you might end up using a version different than the one you expected. Ensure that you set the
QUARTUS_ROOTDIR_OVERRIDE variable after running the setvars script, which can potentially override
your setting.

32
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1

Install Intel® FPGA Board Packages

To compile an executable that can run on an FPGA board, install a Board Support Package (BSP) that allows
targeting compiles to that board. Intel does not ship BSPs with oneAPI. You must download and install BSPs
from a third-party vendor. For more information, refer to the Intel® FPGA development flow page.
To use a third-party vendor-provided BSP with the Intel® oneAPI Base Toolkit, follow these instructions:
1. Follow vendor-specific instructions to download and install the board package.
2. Install a version of the Intel® Quartus® Prime software that is compatible with the BSP (as indicated by
the vendor).
3. Follow instructions in Install the Intel® Quartus® Prime Software to ensure the Intel oneAPI DPC++/C++
Compiler is configured to run with the installed Intel Quartus Prime software.

Related Links
• Intel® oneAPI DPC++/C++ Compiler System Requirements
• Get Started with the Intel® oneAPI Base Toolkit for Windows
• FPGA Flow in the Intel® oneAPI Programming Guide
• FPGA Optimization Guide for Intel® oneAPI Toolkits
• Intel® FPGA SDK for OpenCL™ Pro Edition: Custom Platform Toolkit User Guide
• Intel® Quartus® Prime Software User Guides

Use oneAPI Components in a Yocto Project Build

This section explains how to create a Yocto image with Intel oneAPI components.

System Requirements
Use the Yocto Project official documentation to set up and configure your host machine to make it compatible
with BitBake.

Step 1: Set Up Environment

Set Up Git Repositories
The following Git repositories are required to build a Yocto image:
• Poky
• Meta-intel
• Meta-openembedded
Clone these Git repositories to your host machine:

git clone https://git.yoctoproject.org/git/poky --branch kirkstone

git clone https://git.yoctoproject.org/git/meta-intel --branch kirkstone
git clone https://git.openembedded.org/meta-openembedded --branch kirkstone

Set up BitBake Layers

Execute the following command to set up BitBake layers:

source poky/oe-init-build-env
bitbake-layers add-layer ../meta-intel
bitbake-layers add-layer ../meta-openembedded/meta-oe

Set up BitBake Configurations

33
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Include extra configuration in conf/local.conf` in your build directory as required.

# Set machine
MACHINE = "intel-corei7-64"

# This installs oneAPI packages in the target image.

IMAGE_INSTALL:append = " setup-intel-oneapi-env"
IMAGE_INSTALL:append = " onevpl"
IMAGE_INSTALL:append = " intel-oneapi-compiler intel-oneapi-mkl intel-oneapi-ipp"

Step 2: Build a Yocto Image with oneAPI Packages

Run BitBake to build your image with oneAPI packages. To build the sato image, for example, run:

bitbake core-image-sato

Step 3: Verify the Yocto Image with oneAPI Packages

Verify that oneAPI packages were built successfully. Run the following command:

oe-pkgdata-util list-pkgs | grep intel-oneapi

If the image was built successfully, it will return the list of packages as below:

intel-oneapi-compiler
intel-oneapi-compiler-dbg
intel-oneapi-compiler-dev
intel-oneapi-ipp
intel-oneapi-ipp-dbg
intel-oneapi-ipp-dev
intel-oneapi-mkl
intel-oneapi-mkl-dbg
intel-oneapi-mkl-dev
setup-intel-oneapi-env
setup-intel-oneapi-env-dbg
setup-intel-oneapi-env-dev

Uninstall oneAPI Toolkits and Components

Uninstall Intel PAC Card

To uninstall an installed software stack, you can run one of the following commands:
• Intel® PAC for Intel® Arria® 10 GX FPGA

aocl uninstall intel_a10gx_pac

• Intel® FPGA PAC D5005

aocl uninstall intel_s10sx_pac

After uninstalling the PAC card, you can delete the BSP files and the PAC directory using rm -rf
<pac_install_dir>.

34
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
Uninstall oneAPI Toolkits

NOTE Manual removal of the installation folder (not recommended) does not uninstall the oneAPI
toolkits completely. Incorrect uninstallation may block future installations. If you still need to remove
the toolkit manually, make sure to clean up the system correctly and remove the following:
• Installation folder
• Installer cache, located at /var/intel/installercache/* (root) or <user home>/intel/
installercache/* (user), including:
• packages cache at /var/intel/installercache/packagescache (root) or <user home>/
intel/installercache/packagescache (user)
• download cache at /var/intel/installercache/downloadcache/ (root) or <user home>/
intel/installercache/downloadcache (user)
• (Optional) package manager database /var/intel/packagemanager.db (root) or <user
home>/intel/packagemanager.db (user)
• Installer and package manager executables, located at /opt/intel/oneapi/installer/ (root) or
<user home>/intel/oneapi/installer/ (user) and /opt/intel/packagemanager/ (root) or
<user home>/intel/packagemanager/ (user), respectively.

Uninstall Using GUI

Use the following commands to uninstall oneAPI Toolkits:

cd /opt/intel/oneapi/installer
sudo ./installer
• Select Remove to remove the toolkit
• Use Installer dashboard dialog with the list of already installed products (toolkits) to Modify, Repair or
Remove each toolkit separately

Uninstall Using Silent CLI

Use the following commands to uninstall oneAPI Toolkits:
1. Display the list of already installed products and products included in the downloaded package using the
following command:

l_[Toolkit Name]Kit_[version].sh -s -a --list-products

Example of output:

ID Version Language Installed Name

=================================================================================================
=========
intel.oneapi.lin.tbb.product 2021.1.1-129 false Intel® oneAPI Threading Building Blocks
2. Uninstall selected product:

cd /opt/intel/oneapi/installer
sudo ./installer --action remove --product-id intel.oneapi.lin.tbb.product --product-ver
2021.1.1-129
To uninstall a product from a specific installation instance, use the following command:

cd /opt/intel/oneapi/installer
sudo ./installer --action remove --product-id intel.oneapi.lin.tbb.product --product-ver
2021.1.1-129 --instance=<instance id>

35
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

where <instance id> is a unique combination of alphanumeric symbols set during multi-instance
installation.

Uninstall Using Linux Package Manager

• APT

sudo apt autoremove <package_name>

• YUM

sudo yum autoremove <package_name>

Uninstall PyTorch* and TensorFlow* (Part of Intel® AI Analytics Toolkit)

Uninstall PyTorch
1. Deactivate the pytorch environment, if activated.
2. Uninstall using the following commands:

conda remove -p $<install_dir>/pytorch/1.1.0/ --all

rm -rf $<install_dir>/pytorch

Uninstall TensorFlow
1. Deactivate the tensorflow environment, if activated.
2. Uninstall using the following commands:

conda remove -p $<install_dir>/tensorflow/1.14.0/ --all

rm -rf $<install_dir>/tensorflow

Troubleshooting

Diagnose Errors
The Diagnostics Utility for Intel oneAPI toolkits provides more checks to find missing dependencies and
permissions errors. Learn more.

Integrity check failed

During installation, you may get an error message about failed integrity check of downloaded files.
Cause: Downloaded files are corrupted because of the hard drive corruption.
Solution: Check hard drive health and restart the installation.

Corrupted terminal screen

After launching the installer in CLI mode, you may see a corrupted terminal screen. Dialogs and other
elements are not rendered properly.
Cause: The installer does not support terminal size less than 80x24 characters.
Solution: Resize your terminal size to 80x24 characters or greater before launching the installer.

36
 Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS 1
YUM packages conflict on Amazon Linux 2* OS
When installing toolkits of version 2021.2 or 2021.3 on Amazon Linux 2* OS via YUM, you may get an error
similar to the following:

Error: intel-oneapi-tbb-2021.2.0 conflicts with intel-oneapi-common-

licensing-2021.1.1-2021.1.1-60.noarch
Solution: To work around the issue, install Intel oneAPI toolkits with the following commands:
• Intel® oneAPI Base Toolkit

sudo yum install intel-basekit intel-oneapi-common-licensing-<version> intel-oneapi-libdpstd-

devel-<version>
• Intel® oneAPI HPC Toolkit

sudo yum install intel-hpckit intel-oneapi-common-licensing-<version> intel-oneapi-libdpstd-

devel-<version>
• Intel® oneAPI IoT Toolkit

sudo yum install intel-iotkit intel-oneapi-common-licensing-<version> intel-oneapi-libdpstd-

devel-<version>
• Intel® AI Analytics Toolkit

sudo yum install intel-aikit intel-oneapi-common-licensing-<version>

• Intel® oneAPI Rendering Toolkit

sudo yum install intel-renderkit intel-oneapi-common-licensing-<version>

• Intel® oneAPI DL Framework Developer Toolkit

sudo yum install intel-renderkit intel-oneapi-common-licensing-<version>

where <version> is 2021.2.0 or 2021.3.0. For example:

sudo yum install intel-basekit intel-oneapi-common-licensing-2021.2.0 intel-oneapi-libdpstd-

devel-2021.2.0

2021.x installation overwrites existing 2022.x installer

If you launch the installer for the 2021.x version of a toolkit on a system with the 2022.x version installed,
your existing 2022.x installer will be downgraded to the 2021.x version. The 2021.x installer does not
recognize installed 2022.x packages.
Cause: Compatibility issue between 2022.x and 2021.x versions of the installer.
Solution: Restore the latest installer by launching the installer for the 2022.x version of the toolkit. In
future, when you need to install 2021.x on a system with 2022.x installed, before launching the installer,
back up the following installer directories by renaming them:
• root: /opt/intel/oneapi/installer and /opt/intel/packagemanager/1.0
• user: ~/intel/oneapi/installer and ~/intel/packagemanager/1.0

When you are done with 2021.x, you can change back the directory names to restore the 2022.x installer.

Installation hangs indefinitely when launched in GUI mode using SSH

In an SSH session, installation may hang indefinitely when launched in GUI mode.
Cause: Issue detecting available screen for GUI in an SSH session.
Solution: Interrupt current process using Ctrl + C and launch the installer in CLI mode using the following
command:

sh ./l_[Toolkit Name]Kit_[version].sh -a --cli

37
 1 Intel® oneAPI Toolkits Installation Guide for Linux* OS

Notices and Disclaimers

Intel technologies may require enabled hardware, software or service activation.
No product or component can be absolutely secure.
Your costs and results may vary.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its
subsidiaries. Other names and brands may be claimed as the property of others.

38