Просмотр статьи


Номер журнала: 2019.2

Заголовок статьи: Analysis and optimization of pipelined broadcast algorithms

Резюме

Theoretical and experemental analysis of MPI_Bcast algortihms are performed. The optimal tree degrees and segment sizes for pipelined versions of algorithms are obtained. Algorithms were investigated according to their implementation in the Open MPI. Theoretical results are consistent with experiments on a computer cluster with a Gigabit Ethernet communication network.

Авторы

M. Kurnosov

Ключевые слова

broadcast, MPI, parallel programming, high-performance computing