Sequentialization / Granularity Control [UPM]

• Future plans:
  • Use granularity information to drive data parallelism-like scheduling techniques (e.g., loop unrolling).
  • Extend to Constraint and Concurrent Logic Languages.
  • Study average case analysis (using probability distribution on the inputs). Nontrivial, but profiling techniques can help.
  • Apply results to the distributed versions of ECLPS and Ciao-Prolog.
  • Apply the techniques to other languages.
  • Promising technique for many applications.