TU Darmstadt / ULB / TUbiblio

Automatic Low-Overhead Load-Imbalance Detection in MPI Applications

Arzt, Peter ; Fischler, Yannic ; Lehr, Jan-Patrick ; Bischof, Christian
Sousa, Leonel ; Roma, Nuno ; Tomás, Pedro (eds.) (2021):
Automatic Low-Overhead Load-Imbalance Detection in MPI Applications.
In: Lecture Notes in Computer Science, 12820, In: Euro-Par 2021: Parallel Processing, pp. 19-34,
Cham, Springer International Publishing, Euro-Par 2021: Parallel Processing, Virtual (Lisbon), September 1–3, 2021, ISBN 978-3-030-85665-6,
DOI: 10.1007/978-3-030-85665-6_2,
[Conference or Workshop Item]

Abstract

Load imbalances are a major reason for efficiency loss in highly parallel applications. Hence, their identification is of high relevance in performance analysis and tuning. We present a low-overhead approach to automatically identify load-imbalanced regions and filter out irrelevant ones based on new selection heuristics in our PIRA tool for automatic instrumentation refinement for the Score-P measurement system. For the LULESH mini-app as well as the Ice-sheet and Sea-level System Model simulation package we, thus, correctly identify existing load imbalances while maintaining a runtime overhead of less than 10% for all but one input. Moreover, the traces generated are suitable for Scalasca's automatic trace analysis.

Item Type: Conference or Workshop Item
Erschienen: 2021
Editors: Sousa, Leonel ; Roma, Nuno ; Tomás, Pedro
Creators: Arzt, Peter ; Fischler, Yannic ; Lehr, Jan-Patrick ; Bischof, Christian
Title: Automatic Low-Overhead Load-Imbalance Detection in MPI Applications
Language: English
Abstract:

Load imbalances are a major reason for efficiency loss in highly parallel applications. Hence, their identification is of high relevance in performance analysis and tuning. We present a low-overhead approach to automatically identify load-imbalanced regions and filter out irrelevant ones based on new selection heuristics in our PIRA tool for automatic instrumentation refinement for the Score-P measurement system. For the LULESH mini-app as well as the Ice-sheet and Sea-level System Model simulation package we, thus, correctly identify existing load imbalances while maintaining a runtime overhead of less than 10% for all but one input. Moreover, the traces generated are suitable for Scalasca's automatic trace analysis.

Title of Book: Euro-Par 2021: Parallel Processing
Series Name: Lecture Notes in Computer Science
Volume: 12820
Place of Publication: Cham
Publisher: Springer International Publishing
ISBN: 978-3-030-85665-6
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Scientific Computing
Zentrale Einrichtungen
Zentrale Einrichtungen > University IT-Service and Computing Centre (HRZ)
Zentrale Einrichtungen > University IT-Service and Computing Centre (HRZ) > Hochleistungsrechner
Event Title: Euro-Par 2021: Parallel Processing
Event Location: Virtual (Lisbon)
Event Dates: September 1–3, 2021
Date Deposited: 20 Sep 2021 10:49
DOI: 10.1007/978-3-030-85665-6_2
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)
Show editorial Details Show editorial Details