Logo des Repositoriums

Auflistung nach Autor:in "Kuhn, Michael"

1 - 5 von 5
  • Zeitschriftenartikel
    Automatic Code Transformation of NetCDF Code for I/O Optimisation
    (PARS-Mitteilungen: Vol. 36, 2024) Squar, Jannek; Fuchs, Anna; Kuhn, Michael; Ludwig, Thomas
    Even small improvements to applications can have a huge impact when running on massive parallel systems. Domain experts often lack sufficient computer science expertise or face significant challenges when trying to implement new features such as data compression or parallel I/O. We present anextension to CATO, a code transformation tool that automatically inserts new features and optimisations into scientific code to demonstrate their use and benefits. It helps to overcome initial barriers and supports guided self-learning in a user-friendly way. In this work we implement and evaluate an LLVM pass to automatically find, analyse and transform an application using the netCDF API to optimise the runtime and memory as well as the storage footprint during the I/O phase of the application by inserting parallelisation and compression. Our evaluation shows good speedup and near-optimal memory usage when the modified application is run on distributed hardware using Lustre as the parallel file system backend.
  • Konferenzbeitrag
    Community-Aware Mobile Networking
    (Mobile services and personalized environments – Proceedings of the first international workshop on mobile services and personalized environments, 2006) Kuhn, Michael; Wattenhofer, Roger
    Community related applications such as dating platforms, instant messengers, or file-sharing tools enjoy a great popularity in the Internet. Motivated by this success, community-awareness also has become a hot topic in the mobile world. Despite considerable efforts, however, mobile solutions still cannot compete with their Internet counterparts. A key feature of community-aware applications is profile matching. To a large extent, the missing popularity of mobile approaches can be explained with the problems traditional one-to-one keyword matching causes in a purely decentralized environment: Expressive keywords are unlikely to match because of a lack in user density, and a superficial keyword classification misses the application needs. Hence, we face the challenge to compare characteristic topics in purely mobile applications. To address this challenge, this paper presents a fundamentally new matching scheme based on a combination of topology analysis and distance labeling. Proof-of-concept experiments demonstrate that the proposed scheme has indeed the potential to facilitate the comparison of highly specialized keywords in mobile environments.
  • Conference paper
    Enhancing Parallel Programming Education with High-Performance Clusters Utilizing Performance Analysis
    (Proceedings of DELFI 2024, 2024) Roth, Anna-Lena; James, David; Kuhn, Michael; Konert, Johannes
    The use of high-performance clusters in parallel programming education is not trivial. Not only working on a cluster can be a challenge. Using clusters requires the development of complex parallel programs that are executed with many processes. This increases the necessity for performance analysis to determine bottlenecks, identify optimization opportunities, and evaluate scalability. For learners, performance analysis is a crucial aspect of understanding parallel programs and systems. However, due to its complexity and the prior knowledge required, learners are often overwhelmed by the analysis and interpretation of performance metrics. We conducted a usability test to identify barriers and issues of using a high-performance cluster in parallel programming education in conjunction with professional performance analysis using the tools Score-P, Scalasca, and Cube. At the same time, we tested EduMPI, a novel learning support tool that simplifies the process of executing parallel programs and automates performance analysis at runtime.
  • Zeitschriftenartikel
    Modelling MPI Communication using Coloured Petri Nets
    (PARS-Mitteilungen: Vol. 36, 2024) Krabbe, Tronje; Blesel, Michael; Kuhn, Michael
    The Message Passing Interface (MPI) is a widely used standard for distributed memory, parallel computing. Coloured Petri Nets (CPNs) are a powerful, high-level modelling framework, well suited for modelling distributed systems. This paper presents a novel approach to modelling the communication in MPI programs using Coloured Petri Nets. The paper investigates how this approach can be used for correctness checking of communication schemes. A proof-of-concept software implementation is able to detect several errors and shows promising performance.
  • Zeitschriftenartikel
    Parallele Dateisysteme
    (Informatik Spektrum: Vol. 42, No. 5, 2019) Kuhn, Michael