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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Scaling a parallel program to modern supercomputers is challenging due tointer-process communication, Amdahl's law, and resource contention. Performanceanalysis tools for finding such scaling bottlenecks either base on profiling ortracing. Profiling incurs low overheads but does not capture detaileddependencies needed for root-cause analysis. Tracing collects all informationat prohibitive overheads. In this work, we design ScalAna that uses staticanalysis techniques to achieve the best of both worlds - it enables theanalyzability of traces at a cost similar to profiling. ScalAna first leveragesstatic compiler techniques to build a Program Structure Graph, which recordsthe main computation and communication patterns as well as the program'scontrol structures. At runtime, we adopt lightweight techniques to collectperformance data according to the graph structure and generate a ProgramPerformance Graph. With this graph, we propose a novel approach, calledbacktracking root cause detection, which can automatically and efficientlydetect the root cause of scaling loss. We evaluate ScalAna with realapplications. Results show that our approach can effectively locate the rootcause of scaling loss for real applications and incurs 1.73% overhead onaverage for up to 2,048 processes. We achieve up to 11.11% performanceimprovement by fixing the root causes detected by ScalAna on 2,048 processes.

ScalAna: Automating Scaling Loss Detection with Graph Analysis