A model-based software generation approach qualified for heterogeneous GPGPU-enabled platforms

Holger Endt, Lothar Stolz, Martin Wechs, Walter Stechele

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

An approach for model-based programming is presented supporting automated software generation for Graphics Processing Units (GPUs). As General Purpose Computing on GPUs (GPGPU) has evolved from a very special niche of software technology to an established grade, it is ready now to be adressed from the perspective of automated software generation. This allows to achieve a more flexible, more reliable and faster development process. We present a software generation framework covering the complete design flow from model-based system specification to actual code for heterogeneous CPU and GPU platforms. Good mappings to highly parallel GPU architecture are achieved as inherent parallelism of signal processing algorithms is treated within code translation. First results will be given for an example design from the automotive driver assistance domain.

Original languageEnglish
Title of host publicationApplications, Tools and Techniques on the Road to Exascale Computing
PublisherIOS Press BV
Pages217-223
Number of pages7
ISBN (Print)9781614990406
DOIs
StatePublished - 2012

Publication series

NameAdvances in Parallel Computing
Volume22
ISSN (Print)0927-5452

Keywords

  • ECU Platform
  • GPGPU
  • GPU
  • In-vehicle Driver Assistance
  • Model-Based Programming

Fingerprint

Dive into the research topics of 'A model-based software generation approach qualified for heterogeneous GPGPU-enabled platforms'. Together they form a unique fingerprint.

Cite this