[gecode-users] GeCode vs Minizinc solver

[Christian Schulte]

Dear Andrei,

 

Unfortunately it is not that easy to make Gecode run as fast as the Lazy
solver: they differ essentially in how they do search. 

 

What you could do is try to improve the search annotations in your MiniZinc
model so that you can still use Gecode but with less search.

 

Best

Christian

 

--

Christian Schulte, KTH, web.it.kth.se/~cschulte/

 

From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On Behalf
Of Andrei Neboian
Sent: Monday, April 25, 2011 11:43 AM
To: users
Subject: [gecode-users] GeCode vs Minizinc solver

 

Dear All,

I have built a scheduling model in MiniZinc, which I tried running with
GeCode and the MiniZinc solver.

The in-built LAZY solver of MiniZinc is very efficient and quick in getting
results, while GeCode requires significantly more time to get (first)
results. At the same time, the LAZY solver has a memory issue (and crashes)
when the model is large, but the GeCode doesn't! So I would still like to
try using GeCode instead...

Am I forgetting something when running the GeCode model - why is it so much
slower than the minizinc solver? I do incorporate the GeCode library when
generating the fzn-file out of mzn. Is there anything else? How can I make
GeCode run as fast as the lazy solver of minizinc?

Thanks for help!

Best,
Andi

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gecode.org/pipermail/users/attachments/20110425/03e2ea94/attachment.htm>