On 30 Oct 2012, at 14:42, alexgian wrote: > On 30 October 2012 14:01, Torsten Anders <torsten.and...@beds.ac.uk> wrote: > On 30 Oct 2012, at 11:26, alexgian wrote: >>> > Interesting. I have been meaning to look at gecode for a while, now no >>> > more excuses. What I like best of its features is use of multiprocessors >>> > locally. I am assuming that it will comply with the model described by >>> > Christian Schulte's "Programming Constraint Services" (Well, I see >>> > Christian is heavily involved in gecode). Just wondering, as I see it >>> > has its own search engine. Would that mean there would be two approaches >>> > to search engines, I wonder? >> >> >> Actually, you can do parallel search -- using either multiple local >> processors or networked computers -- already in Mozart 1 (it was broken in >> recent releases introducing the new distribution system, but with older >> versions it is still there). You can see in Christian Schulte's book you >> mentioned that he is actually using Oz (Mozart 1) there ;-) > > Yes, but the idea that multiprocessors will be used in parallel > *tranparently* (I assume) for constraint propagation on the local machine - > whereas I can create parallel searches as I want them for networked computers > - is rather appealing.
I just briefly checked the Gecode documentation. What I understand is that multiple Gecode "workers" (threads), take over different parts of the search tree, but individual spaces (and their propagation) are processed by a single working/thread. Multiple workers can run on different processors of the same machine of multiple networked machines. This is all also the case in Mozart 1 :) Torsten -- Dr Torsten Anders Course Leader, Music Technology University of Bedfordshire Park Square, Room A315 http://www.torsten-anders.de _________________________________________________________________________________ mozart-users mailing list mozart-users@mozart-oz.org http://www.mozart-oz.org/mailman/listinfo/mozart-users
