High-performance PGI Compilers PVFfeatures the latest version of PGIs native OpenMP and auto-parallelFortran compilers. PGI compilers are supported on, and can generatefully optimized code for, a broad range of popular high-performancecomputing platforms including 64-bit x64 multi-core processor-basedsystems as well as 32-bit x86 systems.
In addition to itscomprehensive support for multi-core processors, PGI Visual Fortranincludes PGIs two models for programming GPU accelerators. The PGIAccelerator programming model is a high-level implicit model similar toOpenMP for multi-core x64 systems. PGI Accelerator Fortran enables theincremental offloading of compute-intensive loops and code regions from ahost CPU to a GPU accelerator using simple compiler directives. PGIAccelerator directives are treated as Fortran comments by other Fortrancompilers, so programs incorporating them remain 100% standard-compliantand portable. Developed in cooperation with NVIDIA, CUDA Fortran is ananalog to the NVIDIA CUDA C compiler. CUDA Fortran gives expertprogrammers direct control of all aspects of GPU acceleratorprogramming.