Evaluating the Support of MTC Applications On Intel Xeon Phi Many-Core Accelerators

As Many-Task Computing (MTC) is becoming common-place on clusters, grids, and supercomputers, research that aims to take advantage of the new advances in hardware for MTC workloads is becoming more relevant. A good example is the design of frameworks like GeMTC that incorporate general purpose GPU hardware to improve the concurrency of executing tasks. This work attempts to support MTC workloads on the Intel Xeon Phi accelerators. Our plan is to develop two frameworks that will achieve that goal. One based on OpenMP and the other one based on Intel's Symmetric Communication Interface (SCIF) provided for Many-Integrated Core (MIC) accelerators like the Xeon Phi. Both frameworks aim to provide the same interface as GeMTC, leveraging the integration efforts with the Swift parallel programming system. Our end-goal is to present how programming many-core computing processors can be made easier and more productive using OpenMP or SCIF, and enable the execution of MTC workloads hybrid accelerator-based systems.