Polly-ACC Transparent Compilation to Heterogeneous Hardware

Tobias Grosser, Torsten Hoefler

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Programming today's increasingly complex heterogeneous hardware is difficult, as it commonly requires the use of data-parallel languages, pragma annotations, specialized libraries, or DSL compilers. Adding explicit accelerator support into a larger code base is not only costly, but also introduces additional complexity that hinders long-term maintenance. We propose a new heterogeneous compiler that brings us closer to the dream of automatic accelerator mapping. Starting from a sequential compiler IR, we automatically generate a hybrid executable that - in combination with a new data management system - transparently offloads suitable code regions. Our approach is almost regression free for a wide range of applications while improving a range of compute kernels as well as two full SPEC CPU applications. We expect our work to reduce the initial cost of accelerator usage and to free developer time to investigate algorithmic changes.
Original languageEnglish
Title of host publicationProceedings of the 2016 International Conference on Supercomputing
Place of PublicationNew York, NY, USA
PublisherACM
Pages1:1-1:13
Number of pages13
ISBN (Print)978-1-4503-4361-9
DOIs
Publication statusPublished - 1 Jun 2016
Event2016 International Conference on Supercomputing - Istanbul, Turkey
Duration: 1 Jun 20163 Jun 2016
http://ics16.bilkent.edu.tr/index.htm

Publication series

NameICS
PublisherACM

Conference

Conference2016 International Conference on Supercomputing
Abbreviated titleICS '16
Country/TerritoryTurkey
CityIstanbul
Period1/06/163/06/16
Internet address

Keywords

  • Auto-Parallelization
  • GPGPU
  • Polyhedral Compilation

Fingerprint

Dive into the research topics of 'Polly-ACC Transparent Compilation to Heterogeneous Hardware'. Together they form a unique fingerprint.

Cite this