Learning to Schedule Straight-Line Code

J. Eliot B. Moss, Paul E. Utgoff, John Cavazos, Doina Precup, Darko Stefanovic, Carla E. Brodley, David Scheeff

NeurIPS 1997 pp. 929-935

/neurips/1997/moss1997neurips-learning/

Abstract

Program execution speed on modem computers is sensitive, by a factor of two or more, to the order in which instructions are presented to the proces(cid:173) sor. To realize potential execution efficiency, an optimizing compiler must employ a heuristic algorithm for instruction scheduling. Such algorithms are painstakingly hand-crafted, which is expensive and time-consuming. We show how to cast the instruction scheduling problem as a learning task, ob(cid:173) taining the heuristic scheduling algorithm automatically. Our focus is the narrower problem of scheduling straight-line code (also called basic blocks of instructions). Our empirical results show that just a few features are ad(cid:173) equate for quite good performance at this task for a real modem processor, and that any of several supervised learning methods perform nearly opti(cid:173) mally with respect to the features used.

PDF NeurIPS Semantic Scholar

Cite

Text

Moss et al. "Learning to Schedule Straight-Line Code." Neural Information Processing Systems, 1997.

Markdown

[Moss et al. "Learning to Schedule Straight-Line Code." Neural Information Processing Systems, 1997.](https://mlanthology.org/neurips/1997/moss1997neurips-learning/)

BibTeX

@inproceedings{moss1997neurips-learning,
  title     = {{Learning to Schedule Straight-Line Code}},
  author    = {Moss, J. Eliot B. and Utgoff, Paul E. and Cavazos, John and Precup, Doina and Stefanovic, Darko and Brodley, Carla E. and Scheeff, David},
  booktitle = {Neural Information Processing Systems},
  year      = {1997},
  pages     = {929-935},
  url       = {https://mlanthology.org/neurips/1997/moss1997neurips-learning/}
}