Two techniques for reconciling algorithm parallelism with memory constraints

TitleTwo techniques for reconciling algorithm parallelism with memory constraints
Publication TypeConference Papers
Year of Publication2002
AuthorsVishkin U
Conference NameProceedings of the fourteenth annual ACM symposium on Parallel algorithms and architectures
Date Published2002///
PublisherACM
Conference LocationNew York, NY, USA
ISBN Number1-58113-529-7
Keywordsmemory systems constraints, Parallel algorithms, prefetching
Abstract

The utility of algorithm parallelism for coping with increased processor to memory latencies using "latency hiding" is part of the folklore of parallel computing. Latency hiding techniques increase the traffic to memory and therefore may "hit another wall": limited bandwidth to memory. The current paper attempts to stimulate research in the following general direction: show that algorithm parallelism need not conflict with limited bandwidth.A general technique for using parallel algorithms to enhance serial implementation in the face of processor-memory latency problems is revisited. Two techniques for alleviating memory bandwidth constraints are presented. Both techniques can be incorporated in a compiler.There is often considerable parallelism in many of the algorithms which are known as useful serial algorithms. Interestingly enough, all the examples provided for the use of the two techniques come from such serial algorithms.

URLhttp://doi.acm.org/10.1145/564870.564884
DOI10.1145/564870.564884