Abstract

We describe Moya, an annotation-driven JIT compiler for compiled languages such as Fortran, C and C++. We show that a combination of a small number of easy-to-use annotations coupled with aggressive static analysis that enables dynamic optimization can be used to improve the performance of computationally intensive, long-running numerical applications. We obtain speedups of upto 1.5 on JIT’ed functions and overcome the overheads of the JIT compilation within 25 timesteps in a combustion-simulation application.

Original languageEnglish (US)
Title of host publicationProgramming and Performance Visualization Tools - International Workshops, ESPT 2017 and VPA 2017, Revised Selected Papers
EditorsAbhinav Bhatele, David Boehme, Joshua A. Levine, Allen D. Malony, Martin Schulz
PublisherSpringer
Pages56-73
Number of pages18
ISBN (Print)9783030178710
DOIs
StatePublished - 2019
Event6th Workshop on Extreme-Scale Programming Tools, ESPT 2017 and 4th International Workshop on Visual Performance Analysis, VPA 2017 and Workshop on Extreme-Scale Programming Tools, ESPT 2018 and 5th International Workshop on Visual Performance Analysis, VPA 2018 - Dallas, United States
Duration: Nov 11 2018Nov 16 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11027 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference6th Workshop on Extreme-Scale Programming Tools, ESPT 2017 and 4th International Workshop on Visual Performance Analysis, VPA 2017 and Workshop on Extreme-Scale Programming Tools, ESPT 2018 and 5th International Workshop on Visual Performance Analysis, VPA 2018
Country/TerritoryUnited States
CityDallas
Period11/11/1811/16/18

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Moya—A JIT Compiler for HPC'. Together they form a unique fingerprint.

Cite this