From UNIDAPT.org
SMART'09 Workshop Program
Sunday, January 25, 2009
Paphos, Cyprus
(co-located with HiPEAC 2009 Conference)
Full papers now available! Click on Title
| 14:30-14:40 | SMART'09 Program Chair introduction David Padua, University of Illinois, USA |
| 14:40-16:20 | Session I: Software-oriented approaches Session chair: John Cavazos, University of Delaware, USA |
| 14:40-15:05 | Automatic WCET Reduction by Machine Learning Based Heuristics for Function Inlining Paul Lokuciejewski, Fatih Gedikli, Peter Marwedel, and Katharina Morik TU Dortmund University, Germany |
| 15:05-15:30 | Breaking the Curse of Static Analyses: Making Compiler Intelligent via Machine Learning Lars Alvincz and Sabine Glesner Technical University of Berlin, Germany |
| 15:30-15:55 | Combining Experimental Search and Performance Model for Adaptive Optimization Julien Jaeger* and Denis Barthou*^ * University of Versailles, France ^ INRIA, France |
| 15:55-16:20 | Finding representative sets of optimizations for adaptive multiversioning applications Lianjie Luo*^, Yang Chen*^, Chengyong Wu*, Shun Long** and Grigori Fursin*** * Institute of Computing Technology, China ^ Graduate School of the Chinese Academy of Sciences, China ** JiNan University, China *** INRIA, France |
| 16:20-16:45 | Coffee Break |
| 16:45-18:00 | Session II: Hardware-oriented approaches Session chair: Grigori Fursin, INRIA Saclay, France |
| 16:45-17:10 | Online Linear Regression of Sampling Data from Performance Event Counters Qiang Wu and Oskar Mencer Imperial College London, UK |
| 17:10-17:35 | Using Genetic Programming for Source-Level Data Assignment to Dual Memory Banks Alastair Murray and Björn Franke University of Edinburgh, UK |
| 17:35-18:00 | Adaptive Block Placement Policy for Cache Hierarchies Mohamed Zahran* and Sally A. McKee** * City University of New York, USA ** Chalmers University of Technology, Sweden |
| 18:00-18:10 | Coffee Break |
| 18:10-19:00 | Panel: Can machine learning help to solve the multicore code generation issues? Chair: Francois Bodin, CAPS-Enterprise, France [ Slides ] |
| Compilers are keystone solutions of any approaches that deal with programming homogeneous or heterogeneous multicore architectures. Current general-purpose compilers try to embrace so many domains and try to serve so many constraints that they frequently fail to achieve very high performance. They need to be deeply revisited! In this panel we discuss the use of learning and adaptive techniques in compilers to address the multicore challenges. | |
|
Participants: |