Evolving a CUDA Kernel from an nVidia Template

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

69 Citations (Scopus)

Abstract

Rather than attempting to evolve a complete program from scratch we demonstrate genetic interface programming (GIP) by automatically generating a parallel CUDA kernel with identical functionality to existing highly optimised ancient sequential C code (gzip). Generic GPGPU nVidia kernel C++ code is converted into a BNF grammar. Strongly typed genetic programming uses the BNF to generate compilable and executable graphics card kernels. Their fitness is given by running the population on a GPU with randomised subsets of training data itself derived from gzip's SIR test suite. Back-to-back validation uses the original code as a test oracle.
Original languageEnglish
Title of host publicationUnknown
Place of PublicationNEW YORK
PublisherIEEE
ISBN (Print)978-1-4244-8126-2
Publication statusPublished - 2010
Event2010 IEEE World Congress on Computational Intelligence - Barcelona, Spain
Duration: 18 Jul 201023 Jul 2010

Publication series

Name2010 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION (CEC)

Conference

Conference2010 IEEE World Congress on Computational Intelligence
Country/TerritorySpain
CityBarcelona
Period18/07/201023/07/2010

Fingerprint

Dive into the research topics of 'Evolving a CUDA Kernel from an nVidia Template'. Together they form a unique fingerprint.

Cite this