Presentation 2012-03-02
Profiling-based Source to Source Compiler for GPGPU
Atsushi YUMOTO, Nobuhiko SUGIO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A new profiling-based C to CUDA compiler for GPCPU is proposed in order to help developing higher performance GPGPU applications with less programming effort. The proposed compiler splits a give program into tasks, assigns them into processors with consideration of task characteristics, and finally generates tuned CUDA code, according to profiled parameters. The prototype compiler is implemented by use of our own source-to-source compilation framework called "SCRinfra." Code generation experiment for example programs shows effectiveness of the proposed compiler.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) GPGPU / CUDA / Compiler / Source-to-Source / Accelaration / Profile
Paper # CPSY2011-83,DC2011-87
Date of Issue

Conference Information
Committee DC
Conference Date 2012/2/24(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Dependable Computing (DC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Profiling-based Source to Source Compiler for GPGPU
Sub Title (in English)
Keyword(1) GPGPU
Keyword(2) CUDA
Keyword(3) Compiler
Keyword(4) Source-to-Source
Keyword(5) Accelaration
Keyword(6) Profile
1st Author's Name Atsushi YUMOTO
1st Author's Affiliation Interdisciplinary Graduate School of Science and Engineering Tokyo Institute of Technology()
2nd Author's Name Nobuhiko SUGIO
2nd Author's Affiliation Interdisciplinary Graduate School of Science and Engineering Tokyo Institute of Technology
Date 2012-03-02
Paper # CPSY2011-83,DC2011-87
Volume (vol) vol.111
Number (no) 462
Page pp.pp.-
#Pages 6
Date of Issue