Author: Heiko Mikala
Date: 23:11:10 09/14/98
Go up one level in this thread
On September 15, 1998 at 01:58:33, Kai Skibbe wrote: >On September 14, 1998 at 17:04:40, Heiko Mikala wrote: >> >>Final standings (output using Rebel 8.0): >> >> Name Tot Score + « - Perc Elo >> >>1. WCrafty-15_18 80 63.0 - 17.0 61 4 15 78% 0 >>2. Comet-WB-A90 80 40.5 - 39.5 33 15 32 50% 0 >>3. ZChess 80 37.5 - 42.5 33 9 38 46% 0 >>4. Phalanx-XVII 80 36.0 - 44.0 31 10 39 45% 0 >>5. Goliath105 80 23.0 - 57.0 19 8 53 28% 0 >> > >Hi Heiko, > >I have played several games between Crafty and Phalanx17 and I think >Phalanx has a bug so it will loose many games in a match game (/mg=x). >It has happened that Phalanx scores 6-2 against another engine at blitz level >and than loose the rest of the games. When you take a look at the pgn-file >you see that Phalanx looses in <=40 moves. Sometime <=20 (!!). >Because of this I only play Phalanx with /mg=1 against other machines. So I >get very good results for Phalanx and I think it is very strong. >Could you please test it with the option /mg=1 and post or email me your >experiences ? Hi Kai, yes I found out about the bug too. It happens, whenever Phalanx plays many games in a row, without beeing restartet - for example in match-game mode in Winboard. The bug is, that it will stop thinking, it will play it's moves immediately without using it's time. Because of this I decided to play all Winboard matches with the flags /-reuse and /-reuse2. In this way the engines are restarted for every new game, instead of only starting a new game with the 'new' command using the same engine. I had no problems with Phalanx since then. There should be no 'bad' Phalanx games in my results, because I replayed those first games, in which this phenomenon took place. And yes, I too think, that Phalanx is very strong, I'm looking forward to testing version XVIII ! Heiko.
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.