Lambda the Ultimate

Multiscale Scheduling, Integrating Competitive and Cooperative Parallelism in Theory and in Practice.
G. E. Blelloch, L. Blum, M. Harchol-Balter, and R. Harper. February, 2007.

Our view is that the computing systems of the future exhibit parallelism multiple levels of
granularity, and that these resources must be shared among many users simultaneously. These combination of these two aspects, which we call multiscale parallelism, poses significant challenges for implementing next-generation systems...

We believe that the theoretical issues of multiscale scheduling cannot be properly addressed without carefully considering how these issues will work with particular applications and how they coordinate with the programming languages used to express the parallelism.

This proposed long-term research project (which as far as I can tell was not mentioned here before) is interesting on many levels. From a programming languages perspective the main idea seems to be extending the NESL language, which is based on data parallelism, with such concepts as speculative parallelism that mesh well with multilevel scheduling.