conan.tools.intel

IntelCC

Available since: 1.41.0

This tool helps you to manage the new Intel oneAPI DPC++/C++ and Classic ecosystem in Conan.

Warning

This generator is experimental and subject to breaking changes.

Warning

macOS is not supported for the Intel oneAPI DPC++/C++ (icx/icpx or dpcpp) compilers. For macOS or Xcode support, you’ll have to use the Intel C++ Classic Compiler.

Note

Remember, you need to have installed previously the Intel oneAPI software.

Note

If you are using CMakeToolchain or MSBuildToolchain, you don’t need to use this generator. See intel-cc compiler section for more information.

This generator creates a conanintelsetvars.sh|bat wrapping the Intel script setvars.sh|bat that set the Intel oneAPI environment variables needed. That script is the first step to start using the Intel compilers because it’s setting some important variables in your local environment.

In summary, the IntelCC generator:

  1. Reads your profile [settings] and [conf].
  2. Uses that information to generate a conanintelsetvars.sh|bat script with the command to load the Intel setvars.sh|bat script.
  3. Then, you or the chosen generator will be able to run that script and use any Intel compiler to compile the project.

Note

You can launch the conanintelsetvars.sh|bat before calling your intel compiler to build a project. Also, Conan will automatically call it in the conanfile build(self) method when running any command with self.run.

At first, ensure you are using a profile like this one:

intelprofile
[settings]
...
compiler=intel-cc
compiler.mode=dpcpp
compiler.version=2021.3
compiler.libcxx=libstdc++
build_type=Release
[options]

[build_requires]
[env]
CC=dpcpp
CXX=dpcpp

[conf]
tools.intel:installation_path=/opt/intel/oneapi

The IntelCC generator can be used by name in conanfiles:

conanfile.py
class Pkg(ConanFile):
    generators = "IntelCC"
conanfile.txt
[generators]
IntelCC

And it can also be fully instantiated in the conanfile generate() method:

conanfile.py
from conans import ConanFile
from conan.tools.intel import IntelCC

class App(ConanFile):
    settings = "os", "arch", "compiler", "build_type"

    def generate(self):
        intelcc = IntelCC(self)
        intelcc.generate()

Now, running the command conan install . -pr intelprofile will generate the conanintelsetvars.sh|bat script which will run the Intel setvars script and load all the variables into your local environment.

Custom configurations

You can apply different installation paths and command arguments simply by changing the [conf] entries. For instance:

intelprofile
[settings]
...
compiler=intel-cc
compiler.mode=dpcpp
compiler.version=2021.3
compiler.libcxx=libstdc++
build_type=Release
[options]

[build_requires]
[env]
CC=dpcpp
CXX=dpcpp

[conf]
tools.intel:installation_path=/opt/intel/oneapi
tools.intel:setvars_args=--config="full/path/to/your/config.txt" --force

If we run again a conan install . -pr intelprofile then the conanintelsetvars.sh script (if we are using Linux OS) will contain something like:

conanintelsetvars.sh
. "/opt/intel/oneapi/setvars.sh" --config="full/path/to/your/config.txt" --force

conf

These are the two different entries for IntelCC:

  • tools.intel:installation_path: (required) argument to tells Conan the installation path, if it’s not defined, Conan will try to find it out automatically.
  • tools.intel:setvars_args: (optional) it is used to pass whatever we want as arguments to our setvars.sh|bat file. You can check out all the possible ones from the Intel official documentation.