Author: Uri Blass
Date: 00:06:44 12/24/05
Go up one level in this thread
On December 24, 2005 at 02:45:29, Mike Byrne wrote: >On December 24, 2005 at 02:44:33, Mike Byrne wrote: > >>On December 24, 2005 at 01:53:05, Uri Blass wrote: >> >>>On December 24, 2005 at 01:42:37, Mike Byrne wrote: >>> >>>> Engine Score Ry Ru GF To Cr Za Fr De Sh An SO >>>>He >>>>01 Rybka v1.0 Preview 2 x64 25.5/33 ··· 001 =11 11= 1=0 110 101 11= 111 =11 111 >>>>111 >>>>02 Ruffian 2.1 24.0/33 110 ··· === 01= 1=1 011 1=1 =1= 101 1=1 11= >>>>111 >>>>03 GFruit-4bx 22.5/33 =00 === ··· 110 10= 1== 011 111 ==1 111 =11 >>>>1=1 >>>>04 Toga II 22.0/33 00= 10= 001 ··· 10= 1== =11 110 11= 111 =11 >>>>111 >>>>05 Crafty64 20.0 19.5/33 0=1 0=0 01= 01= ··· 101 101 =11 001 01= 11= >>>>111 >>>>06 Zappa 1.1 18.0/33 001 100 0== 0== 010 ··· 100 1== =11 111 101 >>>>1=1 >>>>07 Fruit 2.1 16.5/33 010 0=0 100 =00 010 011 ··· 001 110 =11 1== >>>>111 >>>>08 Deep Sjeng 1.6 14.5/33 00= =0= 000 001 =00 0== 110 ··· 0== 11= 1=1 >>>>1=1 >>>>09 Shredder 9.1 13.5/33 000 010 ==0 00= 110 =00 001 1== ··· 0=0 111 >>>>110 >>>>10 AnMon 5.50 11.0/33 =00 0=0 000 000 10= 000 =00 00= 1=1 ··· 110 >>>>111 >>>>11 SOS 5.1 for Arena 8.0/33 000 00= =00 =00 00= 010 0== 0=0 000 001 ··· >>>>=11 >>>>12 Hermann 1.5 3.0/33 000 000 0=0 000 000 0=0 000 0=0 001 000 =00 >>>>··· >>>> >>>>198 of 264 games played >>>>Name of the tournament: RoundRobin3 >>>>Site/ Country: WIN-64-MB, United States >>>>Level: Blitz 1/1 >>>>Hardware: Dual AMD Opteron(tm) Processor 244 1804 MHz with 2,046 MB Memory >>>>Operating system: Microsoft Windows Server 2003 Professional Service Pack 1 >>>>(Build 3790) >>>> >>>> Engine Score Ry Ru GF To Cr Za Fr De Sh An SO He >>>>01 Rybka v1.0 P2 x64 25.5 ··· 001 =11 11= 1=0 110 101 11= 111 =11 111 111 >>>>02 Ruffian 2.1 24.0 110 ··· === 01= 1=1 011 1=1 =1= 101 1=1 11= 111 >>>>03 GFruit-4bx 22.5 =00 === ··· 110 10= 1== 011 111 ==1 111 =11 1=1 >>>>04 Toga II 22.0 00= 10= 001 ··· 10= 1== =11 110 11= 111 =11 111 >>>>05 Crafty64 20.0 19.5 0=1 0=0 01= 01= ··· 101 101 =11 001 01= 11= 111 >>>>06 Zappa 1.1 18.0 001 100 0== 0== 010 ··· 100 1== =11 111 101 1=1 >>>>07 Fruit 2.1 16.5 010 0=0 100 =00 010 011 ··· 001 110 =11 1== 111 >>>>08 Deep Sjeng 1.6 14.5 00= =0= 000 001 =00 0== 110 ··· 0== 11= 1=1 1=1 >>>>09 Shredder 9.1 13.5 000 010 ==0 00= 110 =00 001 1== ··· 0=0 111 110 >>>>10 AnMon 5.50 11.0 =00 0=0 000 000 10= 000 =00 00= 1=1 ··· 110 111 >>>>11 SOS 5.1 for Arena 8.0 000 00= =00 =00 00= 010 0== 0=0 000 001 ··· =11 >>>>12 Hermann 1.5 3.0 000 000 0=0 000 000 0=0 000 0=0 001 000 =00 ··· >>> >>>Something in this result seem unlogical and I suspect hardware problem. >>> >>>I do not buy the excuse that it is 1+1 time control because different time >>>control may explain only small difference and the results suggest that ruffian >>>is significantly stronger than shredder at 1+1 time control with ponder on when >>>I know that the opposite is correct at 4 minutes/40 moves ponder off. >>> >>>I think that it may be a good idea to analyze the games later to check if there >>>is an hardware problem when programs were slowed down by a significant factor or >>>a different problem like losing on time or wrong calculation of the results by >>>arena. >>> >>>Note that it is not only ruffian relative to shredder but also zappa relative to >>>shredder and zappa should be relatively weak at fast time control based on the >>>author. >>> >>>Uri >> >>Using Arena 1.1, no hardware problems -- I'm monitoring CPU usage -- no >>abnormalities noted. No one losing on time -- auto flag is off - I turned it >>off becasue both Shredder and SOS had a hard time not getting flagged in game >>1/1 . >> >>Keep in mind I'm using a 64 bit OS -- that may also imapct different programs >>differently. Crafty, Rybka and Zappa are 64 bit compiles. > >Also pondering is off. If pondering is off then the condition is more similiar to CEGT 64 bits help but certainly they cannot explain the failure of shredder relative to ruffian and note that 64 bits cannot improve programs by more than 100 elo even in 1+1 time control and even if you add 100 elo to zappa then it is weaker than shredder based on the tests that I know. Note that except hardware a possible problem that I can think about is that arena may adjudicate games based on score and it adjudicate games that black win as win of white because it may see positive score and translate it to advantage for white. Another possible problem that I can imagine is programs that ponder even when you set ponder to off. 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.