Scalar Memory References in Pipelined Multiprocessors: A Performance Study

Ravi Ganesan, Shlomo Weiss, Shlomo Weiss

Research output: Contribution to journalArticlepeer-review

Abstract

Interleaved memories are essential in pipelined computers to attain high memory bandwidth. As a memory bank is accessed, a reservation is placed on the bank for the duration of the memory cycle, which is often considerably longer than the processor cycle time. This additional parameter, namely, the bank reservation time or the bank busy time, adds to the complexity of the memory model. For Markov models, exact solutions are not feasible even without this additional parameter due to the very large state space of the Markov chain. In this paper we develop a Markov model which explicitly tracks the bank reservation time. Because we only model one processor and the requested bank, the transition probabilities are not known and have to be approximated. The performance predicted by the model is in close agreement with simulation results.

Original languageEnglish
Pages (from-to)78-86
Number of pages9
JournalIEEE Transactions on Software Engineering
Volume18
Issue number1
DOIs
StatePublished - 1992
Externally publishedYes

Keywords

  • Markov chains
  • Pipelined computers
  • interleaved memory
  • memory conflicts
  • supercomputers

Fingerprint

Dive into the research topics of 'Scalar Memory References in Pipelined Multiprocessors: A Performance Study'. Together they form a unique fingerprint.

Cite this