> How hard is it to make a parallel version of GLPK at the moment? It
seems like the new functionality in 4.20 and 4.21 may be directed
towards that? Are there any plans to provide a parallel GLPK?
I did not consider that. However, solution of active subproblems in
the mip solver could be parallelized in a natural way. The problem is
that the standard C has no features for parallel programming.
Andrew Makhorin
For at least some of the search-modes, where one
has a pool of subproblems to be evaluated
independently, one could parallelise in a quite
simple way using OpenMP. OpenMP instructions are
is simply inserted as comments in the C-code and
the compiler, if it supports OpenMP, then takes
care of the parallelisation.
http://www.openmp.org
/Klas
--
==========================================================================
Klas Markström email: [EMAIL PROTECTED]
fax: (+46)90 786 52 22
phone: (+46)90 786 97 21
Department of Mathematics and Mathematical Statistics
Umeå universitet
S-901 87 Umea, Sweden
URL: http://abel.math.umu.se/~klasm/
==========================================================================
_______________________________________________
Help-glpk mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-glpk
