What we contribute is a means for enhancing ease-of-use and performance on mass market systems in the area of large-scale irregular computations. Specifically, we provide a method managing computational, memory and generalized atomic and synchronizing operations and providing a scalable shared global address space, a sequentially consistent programming model for race-free programs, and portability across single and multinode systems built from commodity multicore processors. This method is particularly useful for addressing irregular computations that are too large or demand too much computation time to be solved feasibly on a single conventional computer.