GPU Performance and Power Tuning Using Regression Trees
Open Access
- 11 May 2015
- journal article
- research article
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 12 (2) , 1-26
Abstract
GPU performance and power tuning is difficult, requiring extensive user expertise and time-consuming trial and error. To accelerate design tuning, statistical design space exploration methods have been proposed. This article presents Starchart, a novel design space partitioning tool that uses regression trees to approach GPU tuning problems. Improving on prior work, Starchart offers more automation in identifying key design trade-offs and models design subspaces with distinctly different behaviors. Starchart achieves good model accuracy using very few random samples: less than 0.3% of a given design space; iterative sampling can more quickly target subspaces of interest.Keywords
Funding Information
- DARPA
- National Science Foundation (CCF-0916971)
- C-FAR
- one of six centers of STARnet
- MARCO
This publication has 24 references indexed in Scilit:
- OpenTuner2014
- A large-scale cross-architecture evaluation of thread-coarsening2013
- A performance analysis framework for identifying potential benefits in GPGPU applications2012
- Collective optimizationACM Transactions on Architecture and Code Optimization, 2010
- A Note on Auto-tuning GEMM for GPUs2009
- Program optimization space pruning for a multithreaded gpu2008
- Accurate and efficient regression modeling for microarchitectural performance and power prediction2006
- Compiler optimization-space exploration2003
- Constructing optimal binary decision trees is NP-completeInformation Processing Letters, 1976
- On Grouping for Maximum HomogeneityJournal of the American Statistical Association, 1958