José Manuel Monsalve

Despite the continuous advancement of parallel systems, there remain many open questions regarding programmability, performance portability, and correctness. These challenges become even more daunting when considering the increasing complexity of computational resources: heterogeneity, deeper memory hierarchies, NUMA nodes, larger datasets, and problem complexity. Our research focuses on programming models and paradigms for parallel computing. My interests lie in defining a common abstraction for parallel systems that enables software modularity and portability across current and future architectures. To this end, I explore different programming models for many-core architectures as well as heterogeneous systems containing acceleration devices.