PGI 2017 v17.4 Compiler Released
Yesterday marked an updated release of a community edition build for the NVIDIA-owned PGI code compiler that focuses on code compilation for CPU and GPU execution.
This PGI v17.4 compiler release has a number of bug fixes, improved debugging of C++ on macOS, the initial community edition release for Windows, and the PGI Accelerator Compilers now has atomic add/subtract support of Fortran single precision complex data types, support for CUDA-Fortran wrap-vote operations, updated nvlink, and more.
Those wanting to try out the community/free edition of the NVIDIA PGI C/C++/Fortran code compiler can find the new release here. I'll try to run some fresh benchmarks of it vs. GCC/Clang in the days ahead.
This PGI v17.4 compiler release has a number of bug fixes, improved debugging of C++ on macOS, the initial community edition release for Windows, and the PGI Accelerator Compilers now has atomic add/subtract support of Fortran single precision complex data types, support for CUDA-Fortran wrap-vote operations, updated nvlink, and more.
Those wanting to try out the community/free edition of the NVIDIA PGI C/C++/Fortran code compiler can find the new release here. I'll try to run some fresh benchmarks of it vs. GCC/Clang in the days ahead.
1 Comment