A summary of new features in each version of the PGI 2020 release.

PGI releases version updates, as often as monthly, which may include both new features and bug fixes. The new features added with each version are listed below. Bug fixes are listed on the PGI Release 2020 Technical Problem Reports summary page.

Complete details on the new features listed here are included in the PGI Release Notes for x86-64 and PGI Installation Guide and Release Notes for OpenPOWER.

PGI 2020 version 20.4 (released May 11, 2020)

  • Added support for interoperability with GNU-compiled OpenMP objects.
  • C/C++ Compilers: Added DWARF metadata for inlined function calls.
  • Fortran Compiler:
    • Improvement performance in TRANSFER intrinsic for large contiguous data transfers.
    • Added DWARF metadata for inlined function and subroutine calls.
  • OpenACC and CUDA Fortran:
    • Default CUDA Toolkit version is now 10.2.
    • Updated the implementation of acc_copyin and acc_create to adhere to the OpenACC 2.0 specification.
  • Plus 29 user-requested enhancements and fixes.

PGI 2020 version 20.1 (released February 6, 2020)

See the PGI Release Notes for x86 and PGI Release Notes and Installation Guide for OpenPower for the complete list of new features.

  • OpenACC and CUDA Fortran
    • Added Fortran interfaces to the supplied NVIDIA cuTENSOR library
    • New cutensorEx module maps select Fortran intrinsic functions to the appropriate cuTENSOR functionality.
    • Added support for calling additional built-in CUDA math functions from device code.
    • Added support for capture of the *this pointer by value to the OpenACC C++ compiler.
    • Improved support for attributes(host,device) functions in CUDA Fortran.
    • Implemented support for on_device in CUDA Fortran.
    • Changed copy behavior for OpenACC reductions to adhere to the OpenACC specification.
    • Dropped support for __CUDA_API_VERSION. Use CUDA_VERSION instead.
    • Updated the profiling interface to conform to the OpenACC 3.0 specification.
    • Added support for the CUDA Toolkit version 10.2. CUDA 10.0 is now default.
  • All Compilers
    • Replaced the OpenMP runtime used by the default Linux LLVM compilers.
    • Upgraded all Linux compilers to use LLVM 9.0.
  • Fortran Compiler
    • Improved debugging metadata for select procedures.
    • Improved implementation of the OpenMP API to follow the OpenMP specification.
  • C/C++ Compilers
    • Upgraded the PGI C compiler pgcc significantly.
  • Operating System
    • Added support for recent Linux distributions:
      • CentOS 7.7, 8.0, and 8.1
      • Fedora 30
      • openSUSE Leap 15.1
      • RHEL 7.7, 8.0, and 8.1
      • SLES 15 SP1
      • Ubuntu 19.04
  • Deprecations and Eliminations
    • The pgf77 driver is no longer provided. Use pgfortran to compile F77 Fortran.
    • The PGI Debugger pgdbg has been discontinued.
    • Discontinued PGI Visual Fortran.
    • Dropped support for the macOS platform.
    • Official support for the nonvvm sub-option -ta=tesla and -Mcuda dropped.
    • CUDA 9.2 and 10.0 are no longer included in the PGI packages.
  • Plus 46 user-requested enhancements and fixes.
Click me
Cookie Consent

This site uses cookies to store information on your computer. See our cookie policy for further details on how to block cookies.