Tasking Meets GPUs:Fighting Deadlocks and Other Monsters

Task parallelism is omnipresent these days; whether in data mining or machine learning, for matrix factorization or even molecular dynamics. Despite the successof task parallelism on CPUs, there is currently no performant way to exploit task parallelism of synchronization-critical algorithms on GPUs.Hence, our goal is the development of a task-based programming model to exploit fine-grained task parallelism on heterogeneous hardware