Author: Uri Blass
Date: 23:57:17 12/02/03
Go up one level in this thread
On December 02, 2003 at 15:05:01, Drexel,Michael wrote: >On December 02, 2003 at 09:26:32, Uri Blass wrote: > >>On December 02, 2003 at 08:49:57, Drexel,Michael wrote: >> >>>On December 02, 2003 at 04:44:48, Jouni Uski wrote: >>> >>>>from FCP forum >>>> >>>>Just a litlle tournement with Ruffian 2.0 final version. >>>>30 min/game , AMD XP2800+ @ 2150Mhz >>>>All programs used their own book >>>> >>>> >>>> >>>>ruffian01 Diksmuide 2003 >>>> 1 2 3 >>>>1 Ruffian 2.0.0 **** 0111 ½11½ 6.0/8 >>>>2 Hiarcs 9 1000 **** 101½ 3.5/8 >>>>3 Fritz 8 ½00½ 010½ **** 2.5/8 >>>> >>>> >>>>Jouni >>> >>>I'm not interested at all in biased tourney results. >>> >>>If Ruffian would have scored 2.5 points and Fritz 8 6 points we would never have >>>heared anything of it. >>> >>>Michael >> >>I do not know and I have no idea who made the tournament and if he announced in >>advance that he is going to do tournament with these participants before >>starting. >> >> >>Ruffian scores so far only 4.5/8 in the premier of WBEC >> >>Of course there are not enough games to know but the result so far is bad and >>previous Ruffian scored better in WBEC > >Look at Rebel 12 and The King 3.23 (I suppose default). > >I don't think you can expect it to score 79%. >The participants are very strong winboard engines. > > >> >>For me every result of less than 79% in WBEC suggest that there was not a very >>big improvement in Ruffian after Ruffian1.0.1 already scored 79% against only >>slightly weaker opponents. > >I think they were considerably weaker on average. > >Michael The question is definition of considerably comparing the programs I find 5 programs that played without being updated. 1)Gandalf4.32h(was not updated) 62/100 5.5/8 2)Nimzo2000wb(was not updated) 55.5/100 3.5/8 3)Pharaon2.62(not updated) 54/100 2.5/8 4)Gromit3.8.2(not updated) 47/100 2.5/8 5)Zarkov4.5e(not updated) 49/100 4/8 They did total of 267.5/500 that is 53.5% Today that do so far 18/40 that is 45% If they do less than 43.5% then I agree that there was more than a small improvement in the average level. Another point to consider is the faster hardware of the premier division. It is possible that faster hardware make Ruffian's task harder because it make the opening choice more important relatively to the strength of the programs because there is less hope that the programs will blunder if they get good positions out of book. I remember from peter berger's results that ruffian scored better at longer time control but he compared only time control that were significantly faster than WBEC and it is possible that the optimal time control for Ruffian is not tournament time control but also not fast blitz because fast blitz means that the program cannot take full advantage of better search to outsearch the opponent by more plies when slow time control means that the objective evaluation of the position out of book becomes more important. Uri
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.