Author: Tony Werten
Date: 22:14:51 10/21/01
Go up one level in this thread
On October 21, 2001 at 12:27:18, Antonio Dieguez wrote: >... >>>>Just ask. I'm not a teacher but I don't explaining to interested people. >>> >>>I guess you mean "I don't matter explaining to interested people" :) >> >>Actually I forgot to type in "mind" > >ok :) > >... >>>the exponential ones finished? I meant the search and killers and prunning and >>>stuff, how can you finish that?? since when Xinix outsearch Fritz? :) I think >>>you mean when you just can't advance via this way then try with the other, and >>>change track next time when you have the same problem again. >> >>When you start you can win a lot with killers and pruning and stuff but at a >>certain point the biggest winners are finished and you have to start with the >>then biggest winners wich are smaller than the first ones. > >yep but we could consider the linear improvements(as I called speed) the thing >that most could come to a near finish in comparison with the others. > >>A good evaluation can give a big win, but it's a lot more difficult and takes >>very very much time. > >yea I know! how I know if this thingy is better than that or if this thingy is >worth at all, how I know if incrementing this parameter in 1 or 2 is better, I >don't have infinite computing time. I use a lot of my own common sense to tune >the eval and I don't mind later to verify scientifically if that's better >really. > >Even, amyan has become a bit less materialistic because when tunning the eval I >tend only to increase values, and rarely decrease, no kidding :) > >>>>>By the way, 5 hashtables... wow :) I have only 2, the eval and the normal, by >>>>>the way I have 59% of ram for the eval one, I know that other programs destinate >>>>>much less for that, well, I hope enough ram is used. >>>> >>>>Way too much. ( My experience, not the absolute truth ) >>> >>>may be. I only tried a small set of positions(just 5!) and set it to search to >>>depth 9 and taken the total time used. Next time I will more and more variated >>>positions. I guess this is the way to test this? is a fact that hashtable data >>>affects search in the next move too in a game but the normal and the eval >>>hashtable as well so I hope this cheap way is enough... >> >>Until a certain level, using testpositions is good enough but you need more than >>5. BT2630 is good to start with but you need more than 9 ply for most searches. >>Even then you still have to use common sense. If you know you put in something >>good it can still decrease the testsuite result. > >I though is not necessary to use test sets in wich you must find a move to >measure the % of eval and normal hashtable, because the program will see the >same move at the same depth in 95% of the cases(-5% only because instability!) >so a fixed depth search is not bad. > >>In my program I found a bug that scored a passed pawn on the file it was on >>instead of the rank. I changed it and the testsuite result went down with 50 >>points :( > >that is weird! I thought so as well, at first. But it was quite logical actually. There are a few kingattack positions in BT2630. The point is to sacrify a piece to eliminate the pawnshelter in front of the kingside castled king. Sacrifying this piece also created passed pawns on de f,g or h file, wich my program scored as very advanced passed pawns. Right move, wrong (nonsense) reason. After the change I scored better in endgame positions ( a bit ) but didn't get these attackpositions anymore. Tony > >by te way LCTII seems hard to amyan, lctii says amyan is much weaker than it is >in normal games, so for me that test is no good and uninteresting. A good test >can measure the real strength in games, or in endgames or an aspect etc. > >>Another problem with testsuites is what to do with the results. Suppose you >>change something and 4 positions gain 5 points and 1 looses 30. Total result is >>minus 10. Keep the change in or not ? > >If the scoring is correct then don't keep it. But the problem is the scoring! :) > >I know too this problem, wich is very common for me when trying changes with the >hashtable or killers or the size of the window. > >be well...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.