Co-array Fortran for parallel programming
ACM SIGPLAN Fortran Forum
Defining and Measuring the Productivity of Programming Languages
International Journal of High Performance Computing Applications
Parallel Programmability and the Chapel Language
International Journal of High Performance Computing Applications
CellSs: making it easier to program the cell broadband engine processor
IBM Journal of Research and Development
Rodinia: A benchmark suite for heterogeneous computing
IISWC '09 Proceedings of the 2009 IEEE International Symposium on Workload Characterization (IISWC)
RapidMind: portability across architectures and its limitations
Facing the multicore-challenge
| Hi-index | 0.00 |
Will HPC programmers (have to) adapt to new programming languages and parallelization concepts? Many different languages are currently discussed as complements or successors to the traditional HPC programming paradigm (Fortran/C+MPI). These include both languages designed specifically for the HPC community (e.g. the partitioned global address space (PGAS) languages UPC, CAF, X10 or Chapel) and languages that allow the use of hardware accelerators (e.g. Cn for ClearSpeed accelerator boards, CellSs for IBM CELL and GPGPU languages like CUDA, OpenCL, CAPS hmpp and RapidMind). During the project "Partnership for Advanced Computing in Europe --- Preparatory Phase" (PRACE-PP), developers across Europe have ported three benchmarks to more than 12 different programming languages and assessed both performance and productivity. Their results will help scientific groups to choose the optimal combination of language and hardware to efficiently tackle their scientific problems. This paper describes the framework used for this assessment and the results gathered during the study together with guidelines for interpretation.