MaCS, on its hand, aims at taking advantage of GPI and its programming model, applying some ideas from PaCCS and evaluating different and new options. The implementation of
MaCS is partially based on the observation that a dynamic and asynchronous load balancing scheme, as required by parallel tree search, is orthogonal to the problem being solved. As such,MaCS leverages previous work with the UTS benchmark [1],aimed at characterising load balancing mechanisms, providing yet another use-case for GPI and its programming model,testing if it can be of advantage in this problem domain.