Announcement

Collapse
No announcement yet.

Merging Of Flang/F18 Fortran Compiler Support Into LLVM Has Been Delayed

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Merging Of Flang/F18 Fortran Compiler Support Into LLVM Has Been Delayed

    Phoronix: Merging Of Flang/F18 Fortran Compiler Support Into LLVM Has Been Delayed

    The modern F18/Flang Fortran front-end to LLVM had been set to land in the LLVM mono repository last Monday that could have made it included as part of the LLVM 10.0 branch set for that day. The LLVM 10.0 branching happened as planned but the landing of this Fortran support did not...

    http://www.phoronix.com/scan.php?pag...anding-Delayed

  • #2
    Good, what's the point of 6 months of testing if they dump a huge feature like this in right at the last minute?

    Comment


    • #3
      Is Fortran 2018 any nice?
      I guess that only old people use Fortran and that it is not anything that who haven't used it before learns.

      Is it still that some mathematical libraries are still only available for Fortran?
      Is it still that Fortran is the fastest way to do math?
      I know people use Python (SciPy, NumPy, mathpy) and Julia, as well as GNU Octave, SAGE and Mathematica, but maybe those are all slower?

      Comment


      • #4
        Originally posted by uid313 View Post
        Is Fortran 2018 any nice?
        I guess that only old people use Fortran and that it is not anything that who haven't used it before learns.

        Is it still that some mathematical libraries are still only available for Fortran?
        Is it still that Fortran is the fastest way to do math?
        I know people use Python (SciPy, NumPy, mathpy) and Julia, as well as GNU Octave, SAGE and Mathematica, but maybe those are all slower?
        Most of the tools you mention use a BLAS library (Basic Linear Algebra). The reference implementation of BLAS is written in fortran and other common implementations (e.g. OpenBLAS) have significant fractions of it written in Fortran. So when you are using Numpy, Octave, R, Julia... you are likely using a Fortran library.

        Comment


        • #5
          Originally posted by uid313 View Post
          Is Fortran 2018 any nice?
          I guess that only old people use Fortran and that it is not anything that who haven't used it before learns.

          Is it still that some mathematical libraries are still only available for Fortran?
          Is it still that Fortran is the fastest way to do math?
          I know people use Python (SciPy, NumPy, mathpy) and Julia, as well as GNU Octave, SAGE and Mathematica, but maybe those are all slower?
          Fortran 2018 (along with Fortran >2008) is awesome. High-level just like Python/MATLAB and fast just like C. Hard to find such a combination in any language for scientific computing. Btw, I am not that old. Just a recent PhD graduate in computational sciences.

          Comment


          • #6
            Oh, what a bummer! The addition of the FORTRAN front-end made me consider LLVM 10 as a mandatory update for my development machines (I usually update the compilers every 2-3 years or so). Now I don't know if I'll go to 10.0.0 or to 9.0.1 instead (or even keep my 7.0.0 installation until 11.0.0 is released). But of course I prefer to wait if this means that this new front-end will work better with existing FORTRAN sources. I was so happy with the upcoming 10.0.0!! :-(

            Comment

            Working...
            X