Developer Guide

Developer Guide for Intel® oneAPI Math Kernel Library Linux*

ID 766690
Date 7/13/2023
Public

A newer version of this document is available.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

Document Table of Contents

Using the -qmkl Compiler Option

The Intel® Parallel Studio XE Composer Edition compiler supports the following variants of the -qmkl compiler option:

-qmkl

to link with a certain Intel® oneAPI Math Kernel Library (oneMKL) threading layer depending on the threading option provided:

  • For -qopenmp the OpenMP threading layer for Intel compilers
  • For -tbb the Intel® Threading Building Blocks (Intel® TBB) threading layer


-qmkl=parallel

to link with a certain Intel® oneAPI Math Kernel Library (oneMKL) threading layer depending on the threading option provided:

  • For -qopenmp the OpenMP threading layer for Intel compilers
  • For -tbb the Intel® Threading Building Blocks (Intel® TBB) threading layer

-qmkl=sequential

to link with sequential version of Intel® oneAPI Math Kernel Library (oneMKL).

-qmkl=cluster

to link with Intel® oneAPI Math Kernel Library (oneMKL) cluster components (sequential) that use Intel MPI.

NOTE:

The -qopenmp option has higher priority than -tbbin choosing the Intel® oneAPI Math Kernel Library (oneMKL) threading layer for linking.

For more information on the -qmkl compiler option, see the Intel Compiler User and Reference Guides.

On Intel® 64 architecture systems, for each variant of the -qmkl option, the compiler links your application using the LP64 interface.

If you specify any variant of the -qmkl compiler option, the compiler automatically includes the Intel® oneAPI Math Kernel Library (oneMKL) libraries. In cases not covered by the option, use the Link-line Advisor or see Linking in Detail.