Keynote at SimuTools

Sitemap

Alvin's Homepage  Teaching  2019  RICM4: Probabilité et Simulation  RICM4: Évaluation de performance  2018  RICM4: Probabilité et Simulation  RICM4: Évaluation de performance  M2R Scientific Methodology and Performance Evaluation  2017  RICM4: Probabilité et Simulation  RICM4: Évaluation de performance  M2R Scientific Methodology and Performance Evaluation  2016  RICM4: Probabilité et Simulation  RICM4: Évaluation de performance  M2R Scientific Methodology and Performance Evaluation  M2R Parallel Systems  2015  RICM4: Probabilité et Simulation  RICM4: Évaluation de Performance  M2R Parallel Systems  M2R Scientific Methodology and Performance Evaluation  2014  RICM4: Probabilité et Simulation  RICM4: Évaluation de Performance  M2R Parallel Systems  M2R Scientific Methodology and Performance Evaluation  2013  RICM4: Probabilité et Simulation  RICM4: Évaluation de Performance  M2R Parallel Systems  M2R Performance Evaluation  Evaluation of SimGrid's Lazy Mechanism for Network Settings  2012  Brown Bag Lunch Sessions  M2R Parallel Systems  2011  M2R Parallel Systems  M2R Performance Evaluation  2010  M2R Parallel Systems  M1 MOSIG Operating Systems  2009  M2R Parallel Systems  2008  M2R Parallel Systems  2007  M2R Evaluation de Performance  Research  smpi  Toward Better Simulation of MPI Applications on Ethernet/TCP Networks  Revisiting the LogGPS model for simulation purposes  MESCAL, MOAIS and NANOSIM seminars  A Selection of Links and Articles Related to Reproducible Research  List of Publications Extracted From HAL  Internships  events  PUF/JLPC Summer school on Performance Metrics, Modeling and Simulation of Large HPC Systems  Distributed Lagrangian  misc  Emacs init file written in org-mode  Lab Blog  2016  June  21. Talk on Reproducible Research at the Inria Scientific Days  May  10. Talk at la maison de la simulation  2015  December  03. Talk at R4  August  18. Talk at ILEA  July  01. Joint Laboratory on Exascale Computing  June  15. Presentation of the 1st year Ph.D. students of the LIG  March  25. Intervention ISN  24. Tutorial on Reproducible Research  2014  October  29. A nice trip to Barcelona  June  20. Presentation of the 1st year Ph.D. students of the LIG  12. PUF/JLPC Summer School on Performance Metrics, Modeling and Simulation of Large HPC Systems  02. 4th edition of the SimGrid User Days  May  21. Tutorial on Reproducible Research for the SyncFree Project  15. How to install a decent version of emacs on Mac OS X with org-mode, ess and everything installed  April  22. Tutorial on Reproducible Research  February  28. Journée de formation Bib2HAL  January  22. PIMLIG seminar on the usage of traces in MESCAL, MOAIS, NANOSIM, ...  15. PNG inlining in html files exported from org  09. Local Compiling of Inria Activity Report  2013  September  25. BOINC workshop  July  11. Trying to extract a hierarchical trace for smart data aggregation  June  26. SMPI status report  13. Presentation of the 1st year Ph.D. students of the LIG  07. SONGS and SUD  May  20. BIS Workshop  April  04. Seminar at TU Vienna  03. Paraver Dumb Exporter and SMPI Comparison  March  23. Analyse du temps moyen de réponse dans une file M/GI/1  20. Revisiting Gantt Charts  19. Keynote at ERAD  18. Playing with R and org-mode  14. SMPI Status Report  06. Keynote at SimuTools  2012  September  12. R 101 for MESCAL and MOAIS  January  27. MPI Zoo: message size classification  26. Designed experiments to perform for SMPI modeling  24. SimGrid "easy tasks" or internships  24. Conceptual and Akypuera  23. Uniform designs and linear regression  23. SkaMPI measurements on griffon G5K  20. SKAMPI chez Martin  Agenda

I have been kindly invited to give a keynote at SimuTools. I have presented the efforts of the SimGrid project over the last years to improve the quality of our simulation and methodologies. The pitch was similar to the one of our never-ending article on SimGrid 3. Simulators are often described in term of accuracy, validity, and versatility and since these characteristics are generally seen as "contradictory" and simulation developpers often trade one for the others way too early in the development process. I have briefly explained that hard work and complementary skills (in maths, modeling, system-level programming and even assembly code) could allow to design simulators that are extremely good for all characteristics and provide better results than pre-existing ad hoc simulators. The talk was concluded with discussions on open science and reproductibility issues. I enjoyed it a lot! :)

Here is the final presentation.