Nelson Lossing

Research

My thesis focuses on automatically generate a distributed parallel code from a sequential code.
This code generation is done with simple step-by-step transformations to guarantee the correctness of the result.

I am especially interest on task parallelism that is less known than data parallelism. Task parallelism unrestrict the parallelism on the loop unlike the data parallelism. But the guarantee of all the task has an update data still must be check.

I am part of the PIPS project, a source-to-source compiler made by the CRI of MINES ParisTech with the participations of other universities.

Bibliography

Workshops

• (en) From Data to Effects Dependence Graphs: Source-to-Source Transformations for C
  Nelson Lossing, Pierre Guillou, Mehdi Amini et François Irigoin
  CPC'15, January 2015, London, UK

Seminars

• (en) Semantic analysis for arrays, structures and pointers (slide)
  Nelson Lossing
  7^e Journées de la Compilation, December 2013, Dammarie-les-Lys, FR

Posters

• (en) Compiler Passes for Automatic Task Distribution
  Nelson Lossing
  Progress poster on the 2^nd year of PhD, June 2015

Reports

• (fr) Compilation and Optimisation of Parallel Languges
  Nelson Lossing
  Progress Synopsis on the 2^ndt year of PhD, June 2015
• (fr) Compilation and Optimisation of Parallel Languges
  (fr) Compilation and Optimisation of Parallel Languges (slide)
  Nelson Lossing
  Progress report on the 1^st year of PhD, May 2014
• (fr) Analyse sémantique des tableaux, des structures et des pointeurs
  Nelson Lossing
  MPRI intership report, August 2013
• (en) PIPS : Tutorial to use PIPS with Eclipse
  Nelson Lossing

Contact