Intel® oneAPI Math Kernel Library LAPACK Examples

Developer Reference

  • 2022.1
  • 12/20/2021
  • Public Content
Contents

Compiling and Building Fortran Examples

If possible, compile the Fortran examples with the Fortran 90 compiler, because some examples contain Fortran 90 specific features like the
ADVANCE
 specifier in the
WRITE
 statement.
An LAPACK binary defines the appropriate entry point for the LAPACK routine, which is called from the example according to the default Fortran naming convention. For example, to define the
dgesv
 routine on IA-32 architecture, use
dgesv_
 (lowercase, trailing underscore) on Linux* OS, or
_DGESV
 (uppercase, leading underscore) on Windows* OS.
To compile and build an example:
  1. Run the
    mklvars
     script to set the required environment variables.
  2. Build the application using a Fortran 90 compiler as follows:
    <f90> <compiler options> <example.f> <link line>
    where
    <f90>
     is a Fortran 90 compiler,
    <example.f>
     is any of the oneMKL LAPACK Fortran examples, and
    <link line>
     contains linker options and arguments, including the LAPACK library files required for your application.
    See the following example lines for reference:
    • with Intel® Fortran Compiler, you can use
      ifort -mkl cgesv_ex.f
       on Linux* OS, or
      ifort /Qmkl cgesv_ex.f
       on Windows* OS.
      See the
      Linking Your Application with the Intel oneAPI Math Kernel Library
       section of the oneMKL Developer Guide for more information on linking with the oneMKL.
    • with third-party compilers, you can use
      gfortran -fopenmp -m64 -I${MKLROOT}/include cgesv_ex.f -Wl,--no-as-needed -L${MKLROOT}/lib/intel64 -lmkl_gf_lp64 -lmkl_core -lmkl_gnu_thread -lpthread -lm -ldl
       to compile with the GNU* Fortran compiler on Linux* OS.
      The example above is linking dynamically against 64-bit oneMKL library files. To create a custom link line, use the oneMKL Link Line Advisor or refer to the oneMKL Developer Guide, section
      Linking Your Application with the Intel oneAPI Math Kernel Library
      .

Product and Performance Information

1

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