Performance analysis challenges and framework for high-performance reconfigurable computing

作者: Seth Koehler , John Curreri , Alan D. George

DOI: 10.1016/J.PARCO.2008.01.008

关键词: SpeedupReconfigurable computingHigh-performance reconfigurable computingField-programmable gate arrayVisualizationComputer architectureSoftwareComputer scienceInstrumentation (computer programming)

摘要: Reconfigurable computing (RC) applications employing both microprocessors and FPGAs have potential for large speedup when compared with traditional (software) parallel applications. However, this is marred by the additional complexity of these dual-paradigm systems, making it difficult to identify performance bottlenecks achieve desired performance. Performance analysis concepts tools are well researched widely available but lacking in RC, despite being great importance due applications' increased complexity. In paper, we explore challenges present new techniques automated instrumentation, runtime measurement, visualization RC application behavior. We also ideas integration conventional create a unified tool as our initial framework FPGA instrumentation measurement. Results from case study provided using prototype tool.

参考文章(17)
Ian A. Troxel, Alan D. George, Ryan A. DeVille, Performance Monitoring for Run-time Management of Reconfigurable Devices. ERSA. pp. 175- 181 ,(2005)
Hung-Hsun Su, Dan Bonachea, Adam Leko, Hans Sherburne, Max Billingsley, Alan D. George, GASP! a standardized performance analysis tool interface for global address space programming models parallel computing. pp. 450- 459 ,(2006) , 10.1007/978-3-540-75755-9_54
Barbara Horner-Miller, Proceedings of the 2006 ACM/IEEE conference on Supercomputing ieee international conference on high performance computing data and analytics. ,(2006)
Mateo Valero, Nader Bagherzadeh, Alex Ramirez, Proceedings of the 2nd conference on Computing frontiers CF '05 Computing Frontiers Conference. ,(2005)
Andreas Knüpfer, Bernhard Voigt, Wolfgang E. Nagel, Hartmut Mix, Visualization of repetitive patterns in event traces parallel computing. pp. 430- 439 ,(2006) , 10.1007/978-3-540-75755-9_52
F. Cristian, A probabilistic approach to distributed clock synchronization international conference on distributed computing systems. pp. 288- 296 ,(1989) , 10.1109/ICDCS.1989.37958
Cengiz Erbas, Seyed Sarkeshik, Murat M. Tanik, Different perspectives of the N-Queens problem Proceedings of the 1992 ACM annual conference on Communications - CSC '92. pp. 99- 108 ,(1992) , 10.1145/131214.131227
Kevin Camera, Hayden Kwok-Hay So, Robert W. Brodersen, An integrated debugging environment for reprogrammble hardware systems international symposium on open collaboration. pp. 111- 116 ,(2005) , 10.1145/1085130.1085145
J.L. Tripp, A.A. Hanson, M. Gokhale, H. Mortveit, Partitioning Hardware and Software for Reconfigurable Supercomputing Applications: A Case Study conference on high performance computing (supercomputing). pp. 27- 27 ,(2005) , 10.1109/SC.2005.54
W. Gropp, E. Lusk, A. Chan, M. Snir, D. Wootton, A. Bolmarcich, C.E. Wu, F. Parpia, From Trace Generation to Visualization: A Performance Framework for Distributed Parallel Systems conference on high performance computing (supercomputing). pp. 50- 50 ,(2000) , 10.5555/370049.370458