FAST

fast

“Find a Suitable Topology for Exascale Applications” (FaST) is a research project funded by the German Ministry of Education and Research. It deals with the temporal and spatial placement of processes on high performance computers of the future. It is widely assumed that the current trend in hardware development will continue and that the CPU performance will therefore grow considerably faster than the I/O performance. In order to prevent that these resources become bottlenecks in the system, FaST develops a new scheduling concept which monitors the system resources and locally adapts the distribution of the jobs. For monitoring the system a new agent-based system will be developed. The adaptations to the schedule will be realized by process migration. The effectiveness of the concept will be demonstrated in a prototype implementation using applications like LAMA and mpiBLAST.

Project Partners

  • Johannes Gutenberg University Mainz (Coordinator)
  • Fraunhofer SCAI, Birlinghoven
  • Megware
  • Partec Cluster Competence Center
  • RWTH Aachen
  • TU Munich
  • University of Cologne

Funding Period

01/2014 -- 12/2016

Contact

Prof. Dr.-Ing. André Brinkmann
Dr. Lars Nagel
Dr. Andreas Henkel
Dr. Ramy Gad

Publications

2018

  • Ramy Gad, Simon Pickartz, Tim Süß, Lars Nagel, Stefan Lankes, Antonello Monti, and André Brinkmann. 2018. Zeroing memory deallocator to reduce checkpoint sizes in virtualized HPC environments. The Journal of Supercomputing 74: 6236–6257. DOI

2017

  • Carsten Trinitis, Josef Weidendorfer, and Andre Brinkmann. 2017. Co-Scheduling: Prospects and Challenges. In CO-SCHEDULING OF HPC APPLICATIONS, 1–11. DOI Author/Publisher URL
  • Tim Süß, Nils Döring, Ramy Gad, Lars Nagel, André Brinkmann, Dustin Feld, Eric Schricker, and Thomas Soddemann. 2017. Impact of the scheduling strategy in heterogeneous systems that provide co-scheduling. Seiten: 142–162. Author/Publisher URL

2016

  • Ramy Gad, Simon Pickartz, Tim Süß, Lars Nagel, Stefan Lankes, and André Brinkmann. 2016. Accelerating Application Migration in HPC. In Proceedings of the 11th Workshop on Virtualization in High-Performance Cloud Computing (VHPC) held in conjunction with the International Supercomputing Conference (ISC), 663–673. DOI
  • Tim Süß, Nils Döring, Ramy Gad, Lars Nagel, André Brinkmann, Dustin Feld, Eric Schricker, and Thomas Soddemann. 2016. Impact of the scheduling strategy in heterogeneous systems that provide co-scheduling. Seiten: 37–42. Author/Publisher URL
  • Tim Süß, Nils Döring, Ramy Gad, Lars Nagel, André Brinkmann, Dustin Feld, Thomas Soddemann, and Stefan Lankes. 2016. VarySched : a framework for variable scheduling in heterogeneous environments. Seiten: 489–492. Author/Publisher URL

2014

  • Ramy Gad, Tim Süß, and André Brinkmann. 2014. Compiler driven automatic kernel context migration for heterogeneous computing. Seiten: 389–398. Author/Publisher URL
  • Simon Pickartz, Ramy Gad, Stefan Lankes, Lars Nagel, Tim Süß, André Brinkmann, and Stephan Krempel. 2014. Migration techniques in HPC environments. Seiten: 486–497. Author/Publisher URL