Loading...

Throughput-memory footprint trade-off in synthesis of streaming software on embedded multiprocessors

Hashemi, M ; Sharif University of Technology | 2013

797 Viewed
  1. Type of Document: Article
  2. DOI: 10.1145/2539036.2539042
  3. Publisher: 2013
  4. Abstract:
  5. We study the trade-off between throughput and memory footprint of embedded software that is synthesized from acyclic static dataflow (task graph) specifications targeting distributed memory multiprocessors. We identify iteration overlapping as a knob in the synthesis process by which one can trade application throughput for its memory requirement. Given an initial processor assignment and non-overlapped task schedule, we formally present underlying properties of the problem, such as constraints on a valid iteration overlapping, maximum possible throughput, and minimum memory footprint. Moreover, we develop an effective algorithm for generation of a rich set of design points that provide a range of trade-off options. Experimental results on a number of applications and architectures validate the effectiveness of our approach
  6. Keywords:
  7. Distributed-memory message-passing multiprocessor SoC ; Stream applications ; Distributed-memory multiprocessors ; Effective algorithms ; Embedded multiprocessors ; Iteration overlapping ; Modulo scheduling ; Processor assignments ; Software pipelining ; Stream application ; Embedded systems ; Iterative methods ; Message passing ; Multiprocessing systems ; Throughput ; Multitasking
  8. Source: Transactions on Embedded Computing Systems ; Volume 13, Issue 3 , December , 2013 ; 15399087 (ISSN)
  9. URL: http://dl.acm.org/citation.cfm?id=2539042