J Supercomput (2013) 65:1063–1075 DOI 10.1007/s11227-012-0853-z A preliminary evaluation of OpenACC implementations Ruymán Reyes · Iván López · Juan J. Fumero · Francisco de Sande Published online: 31 January 2013 © Springer Science+Business Media New York 2013 Abstract During the last few years, the availability of hardware accelerators, such as GPUs, has rapidly increased. However, the entry cost to GPU programming is high and requires a considerable porting and tuning effort. Some research groups and ven- dors have made attempts to ease the situation by defining APIs and languages that simplify these tasks. In the wake of the success of OpenMP, industria and academia are working toward defining a new standard of compiler directives to leverage the GPU programming effort. Support from vendors and similarities with the upcoming OpenMP 4.0 standard lead us to believe that OpenACC is a good alternative for devel- opers who want to port existing codes to accelerators. In this paper, we evaluate three OpenACC implementations: two commercial implementations (PGI and CAPS) and our own research implementation, accULL, to evaluate the current status and future directions of the standard. Keywords OpenACC · PGI · HMPP · accULL · Accelerators · GPGPU · CUDA · OpenCL · OpenMP · Compiler · Productivity 1 Introduction The importance of developer productivity in recent years should not be underesti- mated [4]. Much research has been conducted on reducing the programming effort R. Reyes · I. López · J.J. Fumero · F. de Sande () Dept. de Estadística, I. O. y Computación, La Laguna, Spain e-mail: fsande@ull.es R. Reyes e-mail: rreyes@ull.es I. López e-mail: ilopezro@ull.es J.J. Fumero e-mail: jjfumeroa@ull.es