By introducing asynchronous lambdas, many programming languages have leaped ahead in the race for programmable manycore systems, leaving the operating system and its scheduler behind. Instead of hiding application-inherent parallelism behind pools of threads with opaque behavior, asynchronous lambdas allow programmers to explicitly state which parts of a program can be executed in parallel and when this form of parallelism is available. Introducing stretch as a universal performance metric and externalizing part of the lambda-provided knowledge not only to the runtime but also to the operating system scheduler, this paper tries to lay the foundation for OS scheduling to catch up on the road towards heterogeneous elastic manycore systems. © 2014 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Völp, M., & Roitzsch, M. (2014). Elastic manycores * How to bring the os back into the scheduling game? In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8374 LNCS, pp. 749–758). Springer Verlag. https://doi.org/10.1007/978-3-642-54420-0_73
Mendeley helps you to discover research relevant for your work.