High Performance Runtimes for Next Generation Languages

We will attack the implementation gap between new computer architectures and new high productivity computing languages such as X10. We will do this via a three-pronged strategy: (1) prototype new high performance mechanisms; (2) implement a framework for extensibility; and (3) contribute to the design of an entirely new virtual machine. The impact of the shift in multicore architectures is enormous; much has been written in the media. This proposal addresses a key piece of that puzzle---how to best implement new language features designed to address new architectural realities. By addressing this problem, the proposal contributes to solving the largest single technical challenge facing the computer industry today.