The Portland Group, a wholly-owned subsidiary of ST Microelectronics have announced their agreement with AMD to cooperate on the development of compiler technology for AMD FireStream compute accelerators. As part of the agreement, PGI and AMD will investigate and develop technology to enable PGI Fortran and C compilers to generate code directly for AMD FireStream boards and to generate heterogeneous x64+GPU (Graphics Processing Unit) executables that enable automatic utilisation of AMD FireStream products if present in a system.
The Portland Group recently delivered a technology preview of the industry’s first standard Fortran and C compilers that automatically offload computations from an x64 host program to a GPU. Until now, C and C++ developers targeting GPU accelerators have had to rely on language extensions or library-based solutions, and use of GPUs from Fortran applications has been extremely limited.
The PGI x64+GPU programming model enables programmers to accelerate applications simply by adding portable compiler directives, treated as comments by other compilers, to existing standard-compliant Fortran and C programs and recompiling with appropriate compiler options.
This gives individual developers the power to incrementally use accelerator technology where appropriate in large existing applications, eliminating the need for expensive and difficult corporate-level platform migration or acquisition decisions.
Douglas Miles, Director, The Portland Group said that until now, industry CIO’s and government and university lab directors have encountered difficult decisions when evaluating accelerator technologies. Working together with AMD, The Portland Group will work to make FireStream compute accelerators easily accessible to mainstream C and Fortran developers and unlock the power of these devices using PGI compilers that are already in use at most of the significant HPC centers and sites.
Patricia Harrell, Director of Stream Computing, AMD said that The Portland Group clearly understands the needs of today’s HPC developers, and their efforts are instrumental in driving GPU compute acceleration into the mainstream. AMD look forward to working with PGI to help ensure their Fortran and C compilers can take full advantage of the over 1.2 teraFLOPS of single-precision, and over 240 gigaFLOPS of double-precision performance capabilities of the AMD FireStream 9270 compute accelerators.
PGI develops and markets high-performance C/C++ and Fortran compilers and development tools that are widely used by engineers and scientists in high-performance computing (HPC), the field of technical computing engaged in the modeling and simulation of complex processes, such as ocean modeling, weather forecasting, seismic analysis, bioinformatics and other areas.
PGI compilers and tools are designed to extract maximum performance from the latest multi-core x64 processors, including the recently announced 45 nanometer Quad-Core AMD Opteron processors from AMD. PGI compilers are recognised in the HPC community for delivering world-class performance across a wide spectrum of applications and benchmarks, and they are referenced regularly as the industry standard for performance and reliability. Key advantages of PGI compilers and tools can be found on The Portland Group web site.