PROVIDER Β· the veteran language of scientific computing and simulation

Fortran

Fortran, the first widely used high-level language from 1957, is built for arrays and numerics and still powers weather, climate, fluid, and physics simulation on supercomputers with unmatched optimisation and a thin general ecosystem.

Programming Languages & Frameworks International 2026.11.16
Programming Languages & Frameworks International the veteran language of scientific computing and simulation
3.7pts

Pros

4 items
Exceptionally mature optimisation for array and numerical code leading HPC performance
The core codebase of weather, climate, fluid, and physics simulation
Modern Fortran adds modules, OOP, and coarray parallelism
Interoperates with C and is callable from Python through f2py

Cons

4 items
Thin general ecosystem with no web, data, or ML presence
Historical syntax baggage with lagging testing and packaging practice
Talent concentrated in research institutions with narrow hiring
Tooling and IDE support far behind mainstream languages

Product Information

fortran
Category Programming Languages & Frameworks
Company Fortran
Service Region International
Official Website https://fortran-lang.org/
Recommendation Score 3.7 pts
Published 2026-11-16

Detailed Notes

Overview

Fortran, short for Formula Translation, was released by IBM in 1957 as the first widely used high-level language, designed for arrays and numerical work with modern versions from 2008 and 2018 adding modules, object orientation, and coarray parallelism while remaining the mainstay of supercomputing.

For Programming Languages & Frameworks work the value is peak numerical performance with a vast legacy, since weather forecasting, climate modelling, fluid dynamics, structural analysis, and nuclear physics core codes are largely Fortran, compiler optimisation for array operations is exceptionally mature, and it leads HPC benchmarks.

Recognise the boundaries, because the general ecosystem and tooling are thin with almost no use in web, data processing, or machine learning, syntax carries historical baggage such as fixed form and implicit typing with modern practices like testing and packaging lagging, and talent concentrates in research institutions with very narrow hiring.

Core Strengths

  • Exceptionally mature optimisation for array and numerical code leading HPC performance
  • The core codebase of weather, climate, fluid, and physics simulation
  • Modern Fortran adds modules, OOP, and coarray parallelism
  • Interoperates with C and is callable from Python through f2py

Product Ecosystem

Language features

Whole-array operations, modules, derived types, coarrays, and C interop bindings.

Parallelism and HPC

MPI, OpenMP, and GPU offload through OpenACC across supercomputing environments.

Toolchain

gfortran, ifort, or nvfortran compilers with fpm packaging, CMake builds, and f2py bindings.

Drawbacks

  • Thin general ecosystem with no web, data, or ML presence
  • Historical syntax baggage with lagging testing and packaging practice
  • Talent concentrated in research institutions with narrow hiring
  • Tooling and IDE support far behind mainstream languages

Use Cases

  • Numerical simulation and physics modelling (β˜…β˜…β˜…β˜…β˜…) - Supercomputing mainstay
  • Weather and climate modelling (β˜…β˜…β˜…β˜…β˜…) - Foundational code
  • High-performance linear algebra (β˜…β˜…β˜…β˜…β˜…) - Mature optimisation
  • General software development (β˜…) - Not applicable

Pricing

Plan Price Scope
Fortran and gfortran $0 Open source and free
Commercial compilers such as Intel or NVIDIA Bundled licensing Common in HPC
Supercomputer time Institutional or cloud billing The dominant real cost

The language and gfortran are free with Intel or NVIDIA compilers bundled in HPC suites, while the real cost is machine time and people, since Fortran engineers sit in research and commercial teams are hard to staff.

FAQ

  • Is Fortran obsolete? No, weather, physics, and engineering simulation on supercomputers still centre on it, though general software rarely uses it.

  • How does it work with Python? A common pattern writes compute kernels in Fortran called from Python via f2py or C interfaces, balancing productivity and speed.

  • Is modern Fortran worth learning? Only for numerical simulation or inheriting research code, with general development better served by Python, C++, or Rust.

Code Examples

Array maths and formatted output

program plans
    implicit none
    real :: prices(3) = [99.0, 59.0, 129.0]

    print '(A, F6.1)', 'average: ', sum(prices) / size(prices)
    print '(A, F6.1)', 'max: ', maxval(prices)
end program plans

Build with gfortran -O2 -o plans plans.f90 where whole-array maths is Fortran's core strength.

Modules and subroutines

module stats
    implicit none
contains
    pure real function average(x) result(r)
        real, intent(in) :: x(:)
        r = sum(x) / size(x)
    end function average
end module stats

Community & Resources

Communities and official channels

Learning resources and tools

  • Fortran Learn β€” Recommended books and courses
  • fpm β€” Modern build and package tool
  • Fortran Wiki β€” Syntax, compilers, and libraries
Visit Fortran Website β†’ https://fortran-lang.org/
Visit Website β†—