Hallo,
i decieded to release a new unofficial FET version again.
it contain a (hopefully) nice speedup. i already tested very much, but the source is (hopefully not) critical and that is why it will not be included into official version now.
So i hope there will be more testers that help me melt the fear of critical bugs in this speedup idea.
This speedup work to constraints "min n day between activities" with 100% weight. So if your samplefile contain no/less 100% weight min n day constraints it is now a good time to change to 100% and check possible speed improvement. So of course you are allowed to have contraints with less then 100%, but only the constraints with 100% will improve speed.
Don't forget timetabling is a very random process, so please generate at least 20 tables (much more, if you have a difficlt timetable.)
Please let me know your results. I hope generating now will be faster, but if you have an samplefile that generate slower, i am highly ineterssted in that to find my bug.
Please get this new version (windows executable and source):
http://www.timetabling.de/download/unofficial/fet-5.10.2.vd1.exe
i also added a small sample file so you can check speed up very quick (please generate at least 20 files to compare with official version.)
but you don't need to check that samplefile to much, because i already tested it very much. i am much more inetressted on your sample files.
Thank your for your help.
Volker Dirr