MiDataSets for MiBench

Iterative optimization is now a popular technique to obtain performance or code size improvements over the default settings in a compiler. However, in most of the research projects, the best configuration is found for one arbitrary dataset and it is assumed that this configuration will work well with any other dataset that a program uses. We created 20 different datasets per program for MiBench benchmark to evaluate this assumption and analyze the behavior of various programs with multiple datasets. This work has been presented at HiPEAC'07. We expect that MiDataSets will enable research on realistic program optimization, run-time adaptation and benchmarking.

Though we made an effort to include only copyright free datasets from the Internet, mistakes are possible. In such cases, please contact Grigori Fursin as soon as possible and we will try to resolve the issue.

Contributors:
Grigori Fursin(INRIA, France);
Olivier Temam(INRIA, France);
John Cavazos (University of Delaware, USA);
Kenneth Hoste (Ghent University, Belgium);
Veerle Desmet (Ghent University, Belgium);

If you have any questions or comments, please use provided forums or contact Grigori Fursin

`MiDataSets for MiBench` Maintained by [ Alternative SourceForge view ]
News Mailing lists Forums Objectives Downloads Collaborations Documentation and references Links
Objectives:
Iterative optimization is now a popular technique to obtain performance or code size improvements over the default settings in a compiler. However, in most of the research projects, the best configuration is found for one arbitrary dataset and it is assumed that this configuration will work well with any other dataset that a program uses. We created 20 different datasets per program for MiBench benchmark to evaluate this assumption and analyze the behavior of various programs with multiple datasets. This work has been presented at HiPEAC'07. We expect that MiDataSets will enable research on realistic program optimization, run-time adaptation and benchmarking. Though we made an effort to include only copyright free datasets from the Internet, mistakes are possible. In such cases, please contact Grigori Fursin as soon as possible and we will try to resolve the issue. Contributors: Grigori Fursin(INRIA, France); Olivier Temam(INRIA, France); John Cavazos (University of Delaware, USA); Kenneth Hoste (Ghent University, Belgium); Veerle Desmet (Ghent University, Belgium); If you have any questions or comments, please use provided forums or contact Grigori Fursin
Downloads:	Collaborations:
MiDataSets SourceForge releases: [ Link* ]* Related websites: [ MILEPOST GCC - machine learning based research compiler ] [ Continuous Collective Compilation Framework ] [ Interactive Compilation Interface ] [ UNIDAPT Framework - run-time adaptation for statically compiled programs ] [ Previous FCO developments ]	INRIA Saclay , University of Edinburgh Used in MILEPOST project, SARC project, HiPEAC network of excellence
Documentation and references:
Grigori Fursin, Cupertino Miranda, Olivier Temam, Mircea Namolaru, Elad Yom-Tov, Ayal Zaks, Bilha Mendelson, Phil Barnard, Elton Ashton, Eric Courtois, Francois Bodin, Edwin Bonilla, John Thomson, Hugh Leather, Chris Williams, Michael O'Boyle. MILEPOST GCC: machine learning based research compiler. Proceedings of the GCC Developers' Summit, Ottawa, Canada, June 2008 [bib] [pdf] Grigori Fursin, John Cavazos, Michael O’Boyle and Olivier Temam. MiDataSets: Creating The Conditions For A More Realistic Evaluation of Iterative Optimization. Proceedings of the International Conference on High Performance Embedded Architectures & Compilers (HiPEAC 2007), Ghent, Belgium, January 2007 [bib] [pdf]
Links:
Related research (INRIA Saclay, France) UNIDAPT group (INRIA Saclay, France) CARD group (University of Edinburgh, UK) COLO group (University of Edinburgh, UK) MiBench (UMich, USA) PathOpt and PathOpt2 (tool for automated application tuning) - free tool within commercial open-source compiler from QLogic (PathScale) to iteratively search for the best compiler flags. It is easily configurable, works with any languages, does not require project modifications, has many basic search strategies (random, one by one, all but one) and easily extendable. ESTO ( Expert System for Tuning Applications)- iterative optimization tool from IBM to find best compiler flags with genetic algorithms Acovea (Analysis of Compiler Options via Evolutionary Algorithm) - iterative optimization tool to find best compiler flags for C and C++ programs with genetic algorithms PAPI library - performance application programming interface Papiex - command line PAPI measurement tool OProfile - a system profiler for Linux ATLAS FFTW SPIRAL Innovative Computing Laboratory (University of Tennessee, USA) MiBench (UMich, USA) Sandro Bartolini (Universita' degli Studi di Siena, Italy) SPEC (USA) If I forgot a related link or some group, please don't hesitate to contact Grigori Fursin.