CGAL users discussion list

["Laurent Rineau (GeometryFactory)" <>]

On Wednesday 05 May 2010 19:36:46 Edoardo Milotti wrote:
> I have two questions on CGAL developments:
> 
> - I have just installed the latest version of the Intel compiler
> (11.1/088), and it seems that compatibility with CGAL-3.6 is even worse
> than it was before (although the Intel compiler could not compile the
> libraries, it did compile the headers included by my program, so that I
> could compile the program with the icpc and link the gcc-compiled library;
> now this is broken). Are there any plans to include the Intel compiler in
> the list of supported compilers?

I can answer to that question: the Intel Compiler (versions 11.0 and 11.1) is 
tested by CGAL daily testsuite. You must make sure the following compilation 
flags are used:
  -D_GLIBCXX_USE_C99_FP_MACROS_DYNAMIC=1 -fp-model strict -diag-disable 980

(Probably only -fp-model strict is important, and the other flags are there 
to 
silent some warnings.)

*However*, the compilation flag -strict-ansi of the Intel Compiler is (not 
yet) 
supported. I have added that setup (with -strict-ansi), and things are going 
better and better. Probably Intel Compiler 11.1 with -strict-ansi will be 
supported (or almost supported) by CGAL-3.7.

Can you give use an example of the error log you get?

> - I use Delaunay triangulations and alpha shapes with very large sets of
> points (more than one million points), and my code would greatly profit
> from a parallel implementation of the triangulation methods (with OpenMP
> and/or MPI). The paper by Batista et al. "Parallel Multi-Core Geometric
> Algorithms in CGAL" discusses some methods that should be implemented in a
> parallel mode for CGAL: are there plans for a future release?

There are ongoing works:
  http://www.cgal.org/in_progress.html

"Parallel algorithms
Sylvain Pion and Johannes Singler (Karlsruher Institut für Technologie) are 
working on multi-core parallelism of some CGAL algorithms like 
triangulations."

The 3D Delaunay triangulation is the main goal of this work, as far as I 
know, 
and Alpha_shape_3 will probably benefit from that work once it is ready.

But this is not a commitment that the feature will be available for a given 
future release. If Sylvain or Johannes read you mail, you will have more 
recent information.

-- 
Laurent Rineau, PhD
R&D Engineer at GeometryFactory           http://www.geometryfactory.com/
Release Manager of the CGAL Project       http://www.cgal.org/