International audienceTo efficiently exploit high performance computing platforms, applications currently have to express more and more finer-grain parallelism. The OpenMP standard allows programmers to do so since version 3.0 and the introduction of task parallelism. Even if this evolution stands as a necessary step towards scalability over shared memory machines holding hundreds of cores, the current specification of OpenMP lacks ways of expressing dependencies between tasks, forcing programmers to make unnecessary use of synchronization degrading overall performance. This paper introduces libKOMP, an OpenMP runtime system based on the X-Kaapi library that outperforms popular OpenMP implementations on current task-based OpenMP benchmarks, but also provides OpenMP programmers with new ways of expressing data-flow parallelism

Monte-Carlo Tree Search is now a well established algorithm, in games and beyond. We analyze its scalability, and in particular its limitations, and the implications in terms of parallelization, in particular for our program MoGo but also for our Havannah program Shakti. In particular, we get a good efficiency for the parallel versions, both for multicore machines and for message-passing machines, but in spite of promising results in self-play there are situations for which increasing the time per move does not solve anything, and therefore parallelization is not the solution either. Nonetheless, for problems on which the Monte-Carlo part is less biased than in Go, parallelization should be very efficient even without shared memory.