English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The performance of collective operations has been a critical issue since theadvent of MPI. Many algorithms have been proposed for each MPI collectiveoperation but none of them proved optimal in all situations. Differentalgorithms demonstrate superior performance depending on the platform, themessage size, the number of processes, etc. MPI implementations perform theselection of the collective algorithm empirically, executing a simple runtimedecision function. While efficient, this approach does not guarantee theoptimal selection. As a more accurate but equally efficient alternative, theuse of analytical performance models of collective algorithms for the selectionprocess was proposed and studied. Unfortunately, the previous attempts in thisdirection have not been successful. We revisit the analytical model-basedapproach and propose two innovations that significantly improve the selectiveaccuracy of analytical models: (1) We derive analytical models from the codeimplementing the algorithms rather than from their high-level mathematicaldefinitions. This results in more detailed models. (2) We estimate modelparameters separately for each collective algorithm and include the executionof this algorithm in the corresponding communication experiment. Weexperimentally demonstrate the accuracy and efficiency of our approach usingOpen MPI broadcast and gather algorithms and a Grid5000 cluster.

Accurate runtime selection of optimal MPI collective algorithms using analytical performance modelling