A summary of new features in each version of the PGI 2018 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 2018 Technical Problem Reports summary page.
Complete details on the new features listed here are included in the PGI Release Notes for x86-64, PGI Installation Guide and Release Notes for OpenPOWER, and the PGI Visual Fortran® Release Notes.
PGI 2018 version 18.10 (released November 13, 2018)
- PGI Community Edition release
- OpenACC and CUDA Fortran
- Integrated support for the production release of CUDA Toolkit 10.0
- Added support for the use of relocatable device code within shared objects on Linux platforms
- Numerous improvements to OpenACC autocompare feature
- Fortran Compilers
- Added support for Fortran 2018 DO CONCURRENT
- Debug metadata for Fortran COMMON blocks now generated
- Improved MATMUL performance for small matrices
- GCC-compatilbe support for visibility in PGCC and PGC++
- All PGI products for Windows now require Visual Studio 2017
- Plus 42 user-requested enhancements and fixes.
PGI 2018 version 18.7 (released August 16, 2018)
- PGI Accelerator Features and Enhancements
- OpenACC 3.0 true deep copy directives for aggregate data structures in Fortran, C and C++
- PGI Compiler Assisted Software Testing (PCAST) can detect when results diverge between CPU and GPU code versions
- Default CUDA version now matches installed CUDA Driver
- Changed the default NVIDIA GPU compute capability list
- Changed the default size for PGI_ACC_POOL_ALLOC_MINSIZE to 128B
- All PGI Compilers
- LLVM-based code generator for Linux/x86-64 and Linux/OpenPOWER platforms is now based on LLVM 6.0. Performance of generated executables using LLVM-based code generator average 15% faster on several important benchmarks.
- Interoperable with GNU versions up to and including GCC 8.1
- Fortran Compiler
- Fortran 2008 SUBMODULE support
- Assignments to allocatable variables changed to match Fortran 2003 semantics on both host and device
- When using LLVM, debug metadata generated for module variables
- Free-form source lines can now be up to 1000 characters
- All Fortran CHARACTER entities now 64-bit integers
- Improved efficiency of combined "distribute parallel" loop constructs
- DWARF debug information for OpenMP thread private variables
- Changed the output of the pgaccelinfo utility
- Changed the output of the pgcpuid utility
- Dropped support for NVIDIA Fermi GPUs
- Last release to include bundled Microsoft toolchain components
PGI 2018 version 18.5 (released May 31, 2018)
- CUDA Toolkit — Added full support for CUDA 9.2; use the cuda9.2 sub-option with the -ta=tesla or -Mcuda compiler options to compile and link with the integrated CUDA 9.2 toolkit components.
- Added support for Xcode 9.3 on macOS.
- Improved function offset information in runtime tracebacks in optimized (non-debug) modes.
- Plus 10 user-requested enhancements and fixes
PGI 2018 version 18.4 (released April 26, 2018)
- PGI Community Edition release.
- PGI Accelerator Compilers — Added support for CUDA 9.2 if one directs the compiler to a valid installation location of CUDA 9.2 using CUDA_HOME.
- PGI Fortran
- Added support for internal procedures, assigned procedure pointers and internal procedures passed as actual arguments to procedure dummy arguments.
- Added support for intrinsics SCALE, MAXVAL, MINVAL, MAXLOC and MINLOC in initializers.
- Plus 20 user-requested enhancements and fixes
PGI 2018 version 18.3 (released March 19, 2018)
- No now features or capabilities added; bug fixes only.
PGI 2018 version 18.1 (released February 16, 2018)
- Support for Intel Skylake, AMD Zen and IBM POWER9 CPUs
- AVX-512 code generation on the latest Intel Xeon processors
- Full OpenACC 2.6 directives-based parallel programming
- OpenMP 4.5 for x86-64 and OpenPOWER multicore CPUs
- Integrated CUDA 9.1 toolkit and libraries for Tesla GPUs
- Automatic CUDA Fortran kernel schedules
- Partial C++17 support
- GCC 7.2 interoperability
- New PGI fastmath intrinsics library including AVX-512 support
- Optional LLVM-based code generator for Linux x86-64
- Updated MPICH, MVAPICH and Open MPI libraries (pre-compiled)
- Updated NetCDF, parallel NetCDF and ESMF libraries (pre-compiled)
- Enhanced CPU Threads views in the PGI Profiler
- OpenMP Performance Tools (OMPT) interface support
- Support for the latest operating systems
- Updated license management software on x86-64 platforms
- Plus 52 user-requested enhancements and fixes
See also the PGI 2018 New Features summary.