Matthew E. Taylor's Publications

Sorted by DateClassified by Publication TypeSorted by First Author Last NameClassified by Research Category

Evolving Compiler Heuristics to Manage Communication and Contention

Matthew E. Taylor, Katherine E. Coons, Behnam Robatmili, Bertrand A. Maher, Doug Burger, and Kathryn S. McKinley. Evolving Compiler Heuristics to Manage Communication and Contention. In Proceedings of the Twenty-Fourth Conference on Artificial Intelligence (AAAI), July 2010. Nectar Track, 25% acceptance rate
AAAI-2010. This paper is based on results presented in our earlier PACT-08 paper.

Download

[PDF]127.8kB  

Abstract

As computer architectures become increasingly complex, hand-tuningcompiler heuristics becomes increasingly tedious and time consumingfor compiler developers. This paper presents a case study that uses agenetic algorithm to learn a compiler policy. The target policyimplicitly balances communication and contention among processingelements of the TRIPS processor, a physically realized prototype chip.We learn specialized policies for individual programs as well asgeneral policies that work well across all programs. We also employ atwo-stage method that first classifies the code being compiled basedon salient characteristics, and then chooses a specialized policybased on that classification.
This work is particularly interesting for the AI community because it1 emphasizes the need for increased collaboration between AIresearchers and researchers from other branches of computer scienceand 2 discusses a machine learning setup where training on the customhardware requires weeks of training, rather than the more typicalminutes or hours.

BibTeX Entry

@InProceedings(AAAI10-Nectar-taylor,
        author="Matthew E. Taylor and Katherine E. Coons and Behnam Robatmili and Bertrand A. Maher and Doug Burger and Kathryn S. McKinley",
        title="Evolving Compiler Heuristics to Manage Communication and Contention",
        note = "Nectar Track, 25% acceptance rate",
        booktitle="Proceedings of the Twenty-Fourth Conference on Artificial Intelligence ({AAAI})",
        month="July",year="2010", 
        abstract="
As computer architectures become increasingly complex, hand-tuning
compiler heuristics becomes increasingly tedious and time consuming
for compiler developers.  This paper presents a case study that uses a
genetic algorithm to learn a compiler policy.  The target policy
implicitly balances communication and contention among processing
elements of the TRIPS processor, a physically realized prototype chip.
We learn specialized policies for individual programs as well as
general policies that work well across all programs.  We also employ a
two-stage method that first classifies the code being compiled based
on salient characteristics, and then chooses a specialized policy
based on that classification.
<br>
This work is particularly interesting for the AI community because it
1 emphasizes the need for increased collaboration between AI
researchers and researchers from other branches of computer science
and 2 discusses a machine learning setup where training on the custom
hardware requires weeks of training, rather than the more typical
minutes or hours.",
        wwwnote={<a href="http://www.aaai.org/Conferences/AAAI/aaai10.php">AAAI-2010</a>.  This paper is based on results presented in our earlier <a href="b2hd-PACT08-coons.html">PACT-08 paper</a>.},
)

Generated by bib2html.pl (written by Patrick Riley ) on Thu Jul 24, 2014 16:09:10