Author: Mike Byrne
Date: 15:08:48 05/04/04
Go up one level in this thread
On May 04, 2004 at 06:52:01, Uri Blass wrote: >On May 04, 2004 at 06:37:47, Mike Byrne wrote: > >>On May 04, 2004 at 05:27:59, Uri Blass wrote: >> >>>On May 03, 2004 at 22:59:03, Mike Byrne wrote: >>> >>>>I thought I had set 19.12 SE Crafty Standard to plat a game/40 minute match >>>>against Ruffian 2.0.2 >>>> >>>>Wrong _ I had set up Crafty 19.12.SE "Jekyll & Hyatt" (random setting >>>>personality) and to make it more bizare - Jeykll and Hyatt is leading after 21 >>>>games. >>>> >>>> Engine Score Cr Ru S-B >>>>1: Crafty 11.5/21 ····················· 0=00=1=00111==1==1=1= 109.25 >>>>2: Ruffian 2.0.2 9.5/21 1=11=0=11000==0==0=0= ····················· 109.25 >>>> >>>>21 of 100 games played >>>>Name of the tournament: Ruffian Vs Crafty 19.12 Standard G/40 >>>>Site/ Country: , USA >>>>Level: Blitz 40/0 >>>>Hardware: Dual Intel(R) Xeon(TM) CPU 1700MHz 1694 MHz >>>>Operating system: Microsoft Windows 2000 Build 2195 Service Pack 4 >>>> >>>>here are the 7 winning settings >>>> >>>>pers_load 1 838 464 290 278 91 1 5 1 0 0 0 0 40 4 111 114 60 93 84 101 46 37 31 >>>>35 42 30 >>>>pers_load 1 835 514 307 296 90 1 5 1 0 0 0 0 40 -22 95 96 60 100 83 97 58 46 52 >>>>45 33 30 >>>>pers_load 1 936 466 303 320 106 1 5 1 0 0 0 0 40 -10 110 76 60 107 82 102 57 58 >>>>39 49 34 30 >>>>pers_load 1 964 500 316 290 99 1 5 1 0 0 0 0 40 6 100 90 60 102 95 96 53 58 49 >>>>31 57 30 >>>>pers_load 1 918 482 297 276 103 1 5 1 0 0 0 0 40 -24 75 91 60 108 83 103 48 46 >>>>31 53 56 30 >>>>pers_load 1 813 524 278 319 94 1 5 1 0 0 0 0 40 28 76 88 60 98 96 95 47 36 48 56 >>>>36 30 >>>>pers_load 1 860 458 306 296 95 1 5 1 0 0 0 0 40 -58 123 99 60 96 84 104 47 45 30 >>>>30 55 30 >>> >>>7 winning setting? >>>Does it mean that Crafty played with different setting in different games? >> >>Crafty is played with random settings in all 21 games >> >>> >>>What do the numbers mean. >>>I can guess that after 1 you have value for pieces in centi-pawns but I am not >>>sure. >> >>Correct - the rest are various crafty parameters. >> >>Keeper setting 1==yes, 0==no 0 >> queen value..........................925 >> rook value...........................476 >> bishop value.........................276 >> knight value.........................318 >> pawn value...........................99 >> book random (0=off)............ .....1 >> book selection width.................5 >> delay time limit,(0=off:1=on)........1 >> draw score setting...................0 >> search depth.........................0 >> special edition code (1=off:0=on)....0 >> time usage setting (0=normal)........0 >> window beta initial setting..........40 >> >> king safety asymmetry................20 >> king safety scale.............. .....99 >> king safety tropism..................116 >> lazy setting.........................60 >> pass pawn scale......................95 >> pawn block scale............... .....94 >> pawn scale..................... .... 103 >> incheck extension setting............58 >> mate extension setting...............58 >> one response extension setting.......47 >> pawn push extension setting..........42 >> recapture extension setting..........39 >> start extension setting..............30 >> >> >>> >>>I guess that the games were played with ponder on. >>>It may be interesting if you later repeat the experiment but tell Ruffian not to >>>ponder(Crafty can continue to ponder). >> >>I won't have time to that. Perhaps you will. > >No I will not do it for 3 reasons: >1)I do not have dual machine >2)I do not have Ruffian2.02 >3)I do not have computer time to test programs in games unless one of them has >the name movei. > >> >>> >>>I read claims that Ruffian performs worse with pondering and it may be >>>interesting if pondering is not counter productive for it. >>> >>>It is possible that thanks to pondering bug the best setting for Ruffian on dual >>>PC is to tell it not to ponder. >>> >>>If there is no way to tell winboard to tell only one engine to ponder in order >>>to do it then maybe you can change the source code of Crafty to ignore the easy >>>command and ponder. >> >> >>I know nothing of any bug and it was released with a bug , "it is what it is". >>Crafty is ocassionally released with a bug and you see Bob saying don't do this >>or that -- he fixes it. Best solution is to fix the bug and not tell users - do >>not use pondering, do not playgames at xx in yy time etc. Fix the bug, (if >>there is one) is the correct solution. > >I have reasons to suspect that there is a bug based on the material that I read >when Ruffian without pondering scored better. > >I do not tell users what to do. >I only say that I suspect that Ruffian with ponder on is not the strongest >setting for Ruffian. I have reasons to suspect not all chess engines are created equal and some (if not all) have bugs. That's why we run computer vs computer matches - to see (or just gain some insight) which is (may be) the best - bugs and all. You cannot separate the bugs from the engine. If a engine has bug - it's part of the engine -- the same as if engine plays a weak because of poor search , or bad evalaution. So Ruffian or Crafty need not to make any apoligies if one or the other has a bug . That is what computer chess is all about. > >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.