Multiprocessing and Fast Mathematical Capabilities

Numerical codes run fast and efficiently when compiled with Simply Fortran. Our development environment provides the necessary libraries for parallel computing and efficient linear algebra without locking users into a particular CPU architecture or manufacturer. Modern Fortran features, like coarrays, or extensions, like OpenMP, are a simple check box away without any further hassle.

Parallel Computing with OpenMP

Simply Fortran's compiler on Windows includes OpenMP capabilities, and our development environment on GNU/Linux can utilize OpenMP included with the system's Fortran compiler. Parallel computing and other common OpenMP constructs are compiled seamlessly by checking a single project option to enable OpenMP.

Coarray Support

Simply Fortran supports Fortran coarrays on both Windows and GNU/Linux. The Windows compiler includes a shared memory coarray implementation provided by the GNU Fortran compiler, and our Linux package can utilize either OpenCoarrays with Message Passing Interface for older compilers or GNU Fortran's shared memory implementation for newer compilers.

LAPACK and BLAS Included

Simply Fortran includes optimized Linear Algebra Package (LAPACK) and Basic Linear Algebra Subprograms (BLAS) libraries and routines with every installation. These libraries, two of the most common Fortran packages, are easily accessible by enabling a simple project option. Users with code utilizing these common routines can immediately being productive work with Simply Fortran rather than attempting to download, compile, and properly install LAPACK and BLAS.

Additionally, Simply Fortran automatically detects undefined references to LAPACK and BLAS at compile time. If detected, the development environment will suggest and enable linking against LAPACK and BLAS, quickly rectifying this common dependency when working with Fortran code.

Improved Intrinsics on Windows

Our Windows distribution forgoes some custom MinGW mathematical functions for modern, optimized OpenLibm implementations. Our math library is able to take advantage of modern CPU features, leading to noticeably improved efficiency and fewer floating-point exceptions.