Algorithms and implementations for multicore and parallel architectures

Participants

• Sverker Holmgren
• Elisabeth Larsson

• Morgan Ericsson (PostDoc)

• Magnus Gustafsson (PhD student, Principal advisor: S. Holmgren, Co-advisor: M. Thuné)
• Martin Tillenius (PhD student, Principal advisor: E. Larsson, Co-advisor: S. Holmgren)

The research in the group is performed within the Linnaeus Center of Excellence UPMARC. The work e.g. includes collaboration with the Computer Architecture group at the department. Also, the group has contacts with several applied computing groups who contribute with problem settings.

Research

Dynamic Task Scheduling for Molecular Dynamics

A task-based approach with dynamic task scheduling is used to build a molecular dynamics simulation software targeted for multicore systems. Data dependencies are detected by the task library and the tasks are then scheduled dynamically at run-time to fulfill all data dependencies, releasing the programmer from this burden.

The software will be used to generate node point distributions for use in radial basis function methods.

Articles and Refereed Conference Contributions

• Using hardware transactional memory for high-performance computing. Karl Ljungkvist, Martin Tillenius, David Black-Schaffer, Sverker Holmgren, Martin Karlsson, and Elisabeth Larsson. In Proc. 25th International Symposium on Parallel and Distributed Processing Workshops and PhD Forum, pp 1660-1667, IEEE, Piscataway, NJ, 2011. (DOI).
• Implementing Monte Carlo electrostatics simulations on heterogeneous multicore architectures. Marcus Holm, Sverker Holmgren, and Malek Khan. PARA 2010: State of the Art in Scientific and Parallel Computing, University of Iceland, Reykjavík, 2010. (fulltext).
• An efficient task-based approach for solving the <em>n</em>-body problem on multicore architectures. Martin Tillenius and Elisabeth Larsson. PARA 2010: State of the Art in Scientific and Parallel Computing, University of Iceland, Reykjavík, 2010. (fulltext).
• Communication-efficient Krylov methods for exponential integration in quantum dynamics. Magnus Gustafsson, Katharina Kormann, and Sverker Holmgren. PARA 2010: State of the Art in Scientific and Parallel Computing, University of Iceland, Reykjavík, 2010. (fulltext).
• An implementation framework for solving high-dimensional PDEs on massively parallel computers. Magnus Gustafsson and Sverker Holmgren. In Numerical Mathematics and Advanced Applications: 2009, pp 417-424, Springer-Verlag, Berlin, 2010. (DOI).
• Reconsidering algorithms for iterative solvers in the multicore era. Dan Wallin, Henrik Löf, Erik Hagersten, and Sverker Holmgren. In International Journal of Computational Science and Engineering, volume 4, pp 270-282, 2009. (DOI).
• Geographical locality and dynamic data migration for OpenMP implementations of adaptive PDE solvers. Markus Nordén, Henrik Löf, Jarmo Rantakokko, and Sverker Holmgren. In OpenMP Shared Memory Parallel Programming, volume 4315 of Lecture Notes in Computer Science, pp 382-393, Springer-Verlag, Berlin, 2008. (DOI).
• Dynamic data migration for structured AMR solvers. Markus Nordén, Henrik Löf, Jarmo Rantakokko, and Sverker Holmgren. In International journal of parallel programming, volume 35, pp 477-491, 2007. (DOI).

More articles/conf.contr. will be added soon! (2009-08-29)

Technical Reports and Preprints

• Efficient optimization algorithms and implementations for genetic analysis of complex traits on a grid system with multicore nodes. M. Jayawardena, H. Löf, S. Holmgren. Accepted for publication in Proc. Workshop on State-of-the-art in Scientific and Parallel Computing (PARA 2008).

Theses

• Multithreaded PDE Solvers on Non-Uniform Memory Architectures. Markus Nordén. Ph.D. thesis, Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology nr 224, Acta Universitatis Upsaliensis, Uppsala, 2006. (KÖP BOKEN / BUY THIS BOOK, fulltext).
• Iterative and Adaptive PDE Solvers for Shared Memory ArchitecturesIterativa och adaptiva PDE-lösare för parallelldatorer med gemensam minnesorganisation. Henrik Löf. Ph.D. thesis, Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology nr 218, Acta Universitatis Upsaliensis, Uppsala, 2006. (KÖP BOKEN / BUY THIS BOOK, fulltext).
• Methods for Creating and Exploiting Data Locality. Dan Wallin. Ph.D. thesis, Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology nr 176, Acta Universitatis Upsaliensis, Uppsala, 2006. (KÖP BOKEN / BUY THIS BOOK, fulltext).
• Parallelizing the Method of Conjugate Gradients for Shared Memory Architectures. Henrik Löf. Licentiate thesis, IT licentiate theses / Uppsala University, Department of Information Technology nr 2004-005, 2004. (External link).
• Parallel PDE Solvers on cc-NUMA Systems. Markus Nordén. Licentiate thesis, IT licentiate theses / Uppsala University, Department of Information Technology nr 2004-002, 2004. (External link).
• Exploiting Data Locality in Adaptive Architectures. Dan Wallin. Licentiate thesis, IT Licentiate Theses 2003-010, Uppsala: Department of Information Technology, Uppsala University, 2003. (External link).

Conference Contributions

Talks

Posters

Supervised M.Sc. Theses and Project Reports

Software

Contact Information

Sverker Holmgren