Notes on SPC: A Parallel Programming Model

In automatic, retargetable compilation low-cost, analytic cost estimation techniques are crucial in order to e ciently steer the optimization process. Programming models aimed at optimum expressiveness of parallelism, however, are not amenable to static cost estimation. We present a new coordination model, called SPC, that imposes speci c restrictions in the synchronization structures that can be programmed. Imposing these restrictions enables the e cient computation of reliable cost estimations that can be used in the automatic optimization process. We present a conjecture stating that the loss of parallelism due to the synchronization constraints imposed by SPC is limited to a constant factor of 2 compared to the unrestricted case. We demonstrate the expressiveness of SPC through various programming examples, show how SPC is translated, and show how SPC enables automatic program optimizations through a case study compiling a line relaxation kernel on a distributed-memory machine.