In this project, we consider an innovative way to parallelize the execution of large workload applications on a cloud cluster. Instead of parallelizing program fragments using traditional distributed programming like MPI, our framework1 allows the program- mer to abstract the cluster as an OpenMP acceleration device available at the local machine. Although OpenMP has been extensively used to move computation to GPUs, through directive-based annotation, offloading computation to cloud clusters can become a complex and cumbersome task. It typically requires mixing programming models and languages, dealing with various access control mechanisms from different clouds (e.g. AWS and Azure) and integrating all this together into a single application. As a result, this task can become a major programming endeavor that can exclude programmers who are not parallel programming experts from using the computational resources available in the cloud.
H. Yviquel, M. M. Pereira, and G. Araújo, “OmpCloud – Bridging the Gap between OpenMP and Cloud Computing,” in The 2017 OpenMP Developers Conference (Open- MPCon 2017), pp. 3–5, 2017.