PaRA-Sched: a Reconfiguration-Aware Scheduler for Reconfigurable Architectures

被引:12
作者
Cattaneo, Riccardo [1 ]
Bellini, Riccardo [1 ]
Durelli, Gianluca [1 ]
Pilato, Christian [2 ]
Santambrogio, Marco D. [1 ]
Sciuto, Donatella [1 ]
机构
[1] Politecn Milan, Dipartimento Elettron & Informaz, Milan, Italy
[2] Columbia Univ, Dept Comp Sci, New York, NY 10027 USA
来源
PROCEEDINGS OF 2014 IEEE INTERNATIONAL PARALLEL & DISTRIBUTED PROCESSING SYMPOSIUM WORKSHOPS (IPDPSW) | 2014年
关键词
Field Programmable Gate Arrays; High Performance Computing; Reconfigurable Architectures; Scheduling algorithms; Design methodology;
D O I
10.1109/IPDPSW.2014.32
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Harnessing the full capabilities offered by reconfigurable hardware is still a demanding task: the lack of proper methodologies and the intrinsic time consuming and error prone tailoring of these systems around the specific application places a barrier to the adoption of this technology. Partial and Dynamic Reconfiguration (PDR), in this context, is a specific feature whose potential is undiscussed but yet to uncover. In this work, we propose PaRA-Sched, an improvement for a state of the art, highly automated design methodology that allows the designer to rapidly explore the impact of PDR employment during the early stages of the design process. Specifically, we extend the scheduling infrastructure of the framework to explicitly take into account PDR to better explore the design space and improve overall performance by automatically masking reconfiguration time when possible. We show how this additional degree of freedom leads to designs whose performance are improved with respect to the baseline, with a limited increase in time spent during DSE.
引用
收藏
页码:243 / 250
页数:8
相关论文
共 15 条
[1]  
[Anonymous], 2000, PRORISE WORKSH CKTS
[2]  
[Anonymous], 1982, DES AUT C P, DOI DOI 10.1109/DAC.1982.1585498
[3]   PARLGRAN: Parallelism granularity selection for scheduling task chains on dynamically reconfigurable architectures [J].
Banerjee, Sudarshan ;
Bozorgzadeh, Elaheh ;
Dutt, Nikil .
ASP-DAC 2006: 11TH ASIA AND SOUTH PACIFIC DESIGN AUTOMATION CONFERENCE, PROCEEDINGS, 2006, :491-496
[4]   Integrating physical constraints in HW-SW partitioning for architectures with partial dynamic reconfiguration [J].
Banerjee, Sudarshan ;
Bozorgzadeh, Elaheh ;
Dutt, Nikil D. .
IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2006, 14 (11) :1189-1202
[5]   Resource-constrained project scheduling: Notation, classification, models, and methods [J].
Brucker, P ;
Drexl, A ;
Mohring, R ;
Neumann, K ;
Pesch, E .
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 1999, 112 (01) :3-41
[6]  
Clemente J. A., 2011, 2011 International Conference on Field Programmable Logic and Applications, P177, DOI 10.1109/FPL.2011.40
[7]   Optimal FPGA module placement with temporal precedence constraints [J].
Fekete, SP ;
Köhler, E ;
Teich, J .
DESIGN, AUTOMATION AND TEST IN EUROPE, CONFERENCE AND EXHIBITION 2001, PROCEEDINGS, 2001, :658-665
[8]   Ant Colony Heuristic for Mapping and Scheduling Tasks and Communications on Heterogeneous Embedded Systems [J].
Ferrandi, Fabrizio ;
Lanzi, Pier Luca ;
Pilato, Christian ;
Sciuto, Donatella ;
Tumeo, Antonino .
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, 2010, 29 (06) :911-924
[9]   A Design Methodology for Application Partitioning and Architecture Development of Reconfigurable Multiprocessor Systems-on-Chip [J].
Goehringer, Diana ;
Huebner, Michael ;
Benz, Michael ;
Becker, Juergen .
2010 18TH IEEE ANNUAL INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE CUSTOM COMPUTING MACHINES (FCCM 2010), 2010, :259-262
[10]  
Kernighan B.W, 1970, The Bell Systems Technical Journal, V49