Attendees: Antoniu, David, Sebastian, Jan, Tobias, Albert, Ramakrishna, Razya
- Sebastian:
- trunk merged into graphite branch (from now on planned once a week)
- to merge graphite into trunk, we need loop interchange done by end of this week, and we need loop strip mine and blocking by end of June
- PPL folks started to implement the PIP algorithm
- Spec2006 works except dealII (fails with miscompare)
- Tobias:
- discussion on the data dependence analysis for detecting parallel loops
porting GCC to Cloog-ISL and Cloog-PolyLib: translate constraints to polylib matrices, need a polylib-like interface in ISL.
- matrices look like the matrices in scoplib.
- Albert:
- to send out a code for computing the reduced polyhedron for dependences: parametric difference for array dataflow analysis
- Ramakrishna:
- will send to gcc@ list questions about cannonicalization of IVs types
- Razya:
- how to handle reductions for autopar from code marked parallel from graphite? (Discussion on mailing list will be started)
- Li:
Patch to connect autopar with graphite ready. Waiting for DependencyAnalysis.