Networking

TxComm: Transforming Stream Communication for Load Balance, Efficiency, and Fault-Tolerance in Networks-on-Chip

Free registration required

Executive Summary

Recent work has examined using application-specific knowledge of streaming communication to optimize network routing (for throughput/performance) and/or design (for simpler hardware). However, previous techniques have assumed that the communication streams are directly mapped to networks-on-chip. In contrast, this paper explores the use of communication transformations (TxComm) to achieve higher throughput via better network load balance, more efficient network utilization, and better fault-tolerance, while retaining the communication semantics of the original streaming application. Specifically, the authors propose two transformations: stream fission and stream fusion. (While fission and fusion transformations have been applied to computation in streaming programs, they are the first to propose fission and fusion transformations for stream communication.)

  • Format: PDF
  • Size: 331.71 KB