search

akhmadkresna / timetablinghft

Automatically exported from code.google.com/p/timetablinghft
0 stars 0 forks source link

Evaluator performance #7

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
The performance of the evaluator is pretty bad compared to the other systems. 
Right now it is the bottle-neck of the entire system. If some improvement could 
be done here it would clearly boost the algorithm.

Original issue reported on code.google.com by 01weal1...@hft-stuttgart.de on 13 Jun 2010 at 1:30

GoogleCodeExporter commented 9 years ago 
Multithreading would also be an option.

Original comment by 01weal1...@hft-stuttgart.de on 13 Jun 2010 at 2:06

GoogleCodeExporter commented 9 years ago

Original comment by 01weal1...@hft-stuttgart.de on 13 Jun 2010 at 7:10

GoogleCodeExporter commented 9 years ago 
I did some changes and will commit them late in afternoon. maybe comparing the 
evaluator by matthias and me will help...

Original comment by roy...@gmail.com on 14 Jun 2010 at 3:01

GoogleCodeExporter commented 9 years ago 
The new evaluator is almost 3 times as fast:

Maximum Solution Table Size: 50
Iterations: 1000
Reproduction: 85%
Strategy: Course Exchange v8

Old (r214):
comp01.ctt (Duration: 0 h, 0 m, 58 s):

New (r215):
comp01.ctt (Duration: 0 h, 0 m, 19 s):

Original comment by matthias...@gmail.com on 14 Jun 2010 at 5:33

GoogleCodeExporter commented 9 years ago 
Switched to the new evaluator for both performance as well as correctness 
reasons.

Original comment by matthias...@gmail.com on 14 Jun 2010 at 6:36