Fortran + LLVM

llnlGood news everyone! The US government just announced that it has reached an agreement with NVIDIA to produce an open source Fortran front-end for the LLVM compiler infrastructure. It will be based on the existing commercial Portland Group compiler (NVIDIA purchased the Portland Group a couple of years ago). Source code for the Fortran front-end is expected to be available in late 2016. From the announcement:

The project is being spearheaded by the Lawrence Livermore, Sandia and Los Alamos national laboratories in response to the need for a robust open-source Fortran solution to complement and support the burgeoning use of LLVM and the CLANG C++ compiler in the HPC community. Large HPC applications, such as those developed by the NNSA Laboratories, are often built on mixed-language modules, and require a common compiler infrastructure that supports both C/C++ and Fortran. Fortran also remains widely used in the broader scientific computing community, supporting simulation science to advance national security, medicine, energy, climate and basic science missions.

Does Fortran support for LLVM mean that we’ll eventually be able to have Fortran code running on our iPads? Only time will tell…

See also

Posted in Programming Tagged with: , ,
One comment on “Fortran + LLVM
  1. Stefano says:

    This a very good news, however I have had bad experience with PGI. Some years ago I finally proved to my managers that Intel and GNU gfortran perform much better than PGI, especially when dealing with pointers or allocatable components into derived types. Indeed, another “key-feature” was level of warn/debug offered: Intel and GNU gfortran were able to catch much more non-standard or bad-practice programming.

    Some years ago I see the wonderful work that PGI did in supporting almost “natively” the GPUs power embracing CUDA into Fortran. This was a valuable effort in theory, but a lot of “cons” were under the wood. This approach was not standard, thus porting large projects to PGI-CUDA had the risk to chain definitively to a commercial company that tomorrow could decide to stop the support to CUDA… I prefer to rely on standard whenever is possible, thus I am now very excited about Fortran Coarrays.

    I hope that PGI Fortran is now better than it was years ago, and I am very happy to see it open-source for LLVM compiler collection.

    My best regards,
    Stefano

Leave a Reply

Your email address will not be published. Required fields are marked *

*