CUDA-Lite: Reducing GPU programming complexity

Sain Zee Ueng, Melvin Lathara, Sara S. Baghsorkhi, Wen Mei W. Hwu

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

Abstract

The computer industry has transitioned into multi-core and many-core parallel systems. The CUDA programming environment from NVIDIA is an attempt to make programming many-core GPUs more accessible to programmers. However, there are still many burdens placed upon the programmer to maximize performance when using CUDA. One such burden is dealing with the complex memory hierarchy. Efficient and correct usage of the various memories is essential, making a difference of 2-17x in performance. Currently, the task of determining the appropriate memory to use and the coding of data transfer between memories is still left to the programmer. We believe that this task can be better performed by automated tools. We present CUDA-lite, an enhancement to CUDA, as one such tool. We leverage programmer knowledge via annotations to perform transformations and show preliminary results that indicate auto-generated code can have performance comparable to hand coding.

Original languageEnglish (US)
Title of host publicationLanguages and Compilers for Parallel Computing - 21st International Workshop, LCPC 2008, Revised Selected Papers
Pages1-15
Number of pages15
DOIs
StatePublished - 2008
Event21st International Workshop on Languages and Compilers for Parallel Computing, LCPC 2008 - Edmonton, AB, Canada
Duration: Jul 31 2008Aug 2 2008

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5335 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other21st International Workshop on Languages and Compilers for Parallel Computing, LCPC 2008
Country/TerritoryCanada
CityEdmonton, AB
Period7/31/088/2/08

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'CUDA-Lite: Reducing GPU programming complexity'. Together they form a unique fingerprint.

Cite this