Hello Marc, your mail had to be forwarded manually because you are not subcribed to the GLPK bug list, see https://lists.gnu.org/mailman/listinfo/bug-glpk
Are solving the same problem in each round? What time does glpsol need to solve your second problem? Best regards Xypron On 30.10.2012 20:20, Andrew Makhorin wrote: > -------- Forwarded Message -------- > Subject: Inconsistent run times > Date: Tue, 30 Oct 2012 14:34:06 -0400 > > I am running glpk 4.44 downloaded from sourgeforge for 64-bit windows > I am writing a problem formulation to an LP file and then reading it > into glpsol and also read it from my compiled program. > I am using Visual Studio 2010 to compile my 64-bit program. > I have observed the following inconsistencies > > The first time a problem is imported and solved (it is a mip > formulation) it runs about as fast as using glpsol. > The following times, the run times are 10 to 12 times LONGER (e.g. 1.2 > seconds first time, 14.2 seconds thereafter) > In my program I am timing only the mipopt call, not reading the lp file. > > Sequence of calls > > create problem > read lp file > mipopt with presolve > destroy problem > > repeat the above sequence > > The problem is destroyed and the corresponding glpk pointer is null > between iterations. > However the glpk library is not unloaded between repeats of this sequence > When my program is restarted the sequence of events repeats, i.e. only > the first time are the run times short. > > Formulation is ultimately solved and no error codes are return (i.e. > solution is reported as primal optimal which it is) > > I have repeated this behavior on several different Windows 7 machines. > > > > > > > > _______________________________________________ > Bug-glpk mailing list > [email protected] > https://lists.gnu.org/mailman/listinfo/bug-glpk > _______________________________________________ Bug-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/bug-glpk
