TU Darmstadt / ULB / TUbiblio

PIRA: Performance Instrumentation Refinement Automation

Lehr, Jan-Patrick and Hück, Alexander and Bischof, Christian (2018):
PIRA: Performance Instrumentation Refinement Automation.
New York, NY, USA, ACM, In: Proceedings of the 5th ACM SIGPLAN International Workshop on Artificial Intelligence and Empirical Methods for Software Engineering and Parallel Computing Systems, Boston, MA, USA, In: AI-SEPS 2018, ISBN 978-1-4503-6067-8,
DOI: 10.1145/3281070.3281071, [Online-Edition: http://doi.acm.org/10.1145/3281070.3281071],
[Conference or Workshop Item]

Abstract

In this paper we present PIRA – an infrastructure for automatic instrumentation refinement for performance analysis. It automates the generation of an initial performance overview measurement and gradually refines it, based on the recorded runtime information. This can help a performance analyst with the time consuming and largely manual, yet mechanical, task of selecting which functions to capture in subsequent measurements. PIRA implements an existing aggregation strategy that heuristically determines which functions to include for initial overview measurements. Moreover, it implements a newly developed heuristic to incorporate profile information and expand instrumentation in hot-spot regions only. The approach is evaluated on different benchmarks, including the SU 2 multi-physics solver package. PIRA is able to generate instrumentation configurations that contain the application’s hot-spot, but generate significantly less overhead when compared to the Score-P reference measurement.

Item Type: Conference or Workshop Item
Erschienen: 2018
Creators: Lehr, Jan-Patrick and Hück, Alexander and Bischof, Christian
Title: PIRA: Performance Instrumentation Refinement Automation
Language: English
Abstract:

In this paper we present PIRA – an infrastructure for automatic instrumentation refinement for performance analysis. It automates the generation of an initial performance overview measurement and gradually refines it, based on the recorded runtime information. This can help a performance analyst with the time consuming and largely manual, yet mechanical, task of selecting which functions to capture in subsequent measurements. PIRA implements an existing aggregation strategy that heuristically determines which functions to include for initial overview measurements. Moreover, it implements a newly developed heuristic to incorporate profile information and expand instrumentation in hot-spot regions only. The approach is evaluated on different benchmarks, including the SU 2 multi-physics solver package. PIRA is able to generate instrumentation configurations that contain the application’s hot-spot, but generate significantly less overhead when compared to the Score-P reference measurement.

Series Name: AI-SEPS 2018
Place of Publication: New York, NY, USA
Publisher: ACM
ISBN: 978-1-4503-6067-8
Uncontrolled Keywords: Score-P, automatic program instrumentation, high-performance computing, performance engineering
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Scientific Computing
Exzellenzinitiative
Exzellenzinitiative > Graduate Schools
Exzellenzinitiative > Graduate Schools > Graduate School of Computational Engineering (CE)
Zentrale Einrichtungen
Zentrale Einrichtungen > University IT-Service and Computing Centre (HRZ)
Zentrale Einrichtungen > University IT-Service and Computing Centre (HRZ) > Hochleistungsrechner
Event Title: Proceedings of the 5th ACM SIGPLAN International Workshop on Artificial Intelligence and Empirical Methods for Software Engineering and Parallel Computing Systems
Event Location: Boston, MA, USA
Date Deposited: 26 Nov 2018 14:29
DOI: 10.1145/3281070.3281071
Official URL: http://doi.acm.org/10.1145/3281070.3281071
Export:

Optionen (nur für Redakteure)

View Item View Item