Abstract

The rise of commodity multicore processors makes parallel computing available to the masses. Traditional parallel languages focus on large-scale scientific computing and are not well suited to programming the applications one typically finds on desktop systems. Such desktop applications are better supported by heterogeneous parallel languages that provide a spectrum of parallel constructs working at different granularities. In this paper, we focus on the problem of how to support a heterogeneous collection of parallel-programming mechanisms in a compiler and runtime system. We take a micro-kernel approach in our design: the compiler and runtime support a small collection of scheduling primitives upon which complex scheduling policies can be implemented. Our approach is part of a larger effort to design and implement a parallel functional programming language, but it is flexible enough to support a wide range of possible parallel-programming mechanisms. We give examples of a number of different schedulers, provide a formal specification of the runtime model, and describe our implementation. 1.

Citations