Firebird 1.0 - Rybka 3 ,
Firebird holds his superiority at longer timecontrols in serie of Nunn2 testmatches
The ELO difference between the different engines and Rybka, tested on CCRL and CEGT, at short (40/4) and longer (40/20 an 40/120) timecontrols, are pretty consistent. Which indicates that when there is a substantial difference - e.g. > 60 Elo - at a shorter timecontrol, than there will be also a sustained difference at longer timecontrol.
But most remarkably, almost all of the engines perform better against Rybka at longer timecontrols. So there is a tendency that the difference between a program and number one Rybka, becomes clearly less at longer timecontrols. This is most obviously the case for Naum and Zappa.
My point, there is no evidence to believe that this will not be the case with the difference between Firebird 1.0 beta and Rybka 3 at larger timecontrols.
CEGT - MP versions x64, 4CPU at 40/4; 40/20 and 40/120
Rybka Stockfish Naum4 Deep Fritz11 Zappa Mexico II Fruit 2.4 betaA
40/4 3236 3147 3123 3065 3023 2983
diff with rybka 89 113 171 213 253
40/20 3188 3127 3094 3020 3017 2952
diff with rybka 61 94 168 171 236
40/120 3166 ? 3105 3014 3003 2950
diff with rybka ? 61 152 163 216
My testing over 2x 50 games Nunn2 testmatch, between Firebird 1.0 beta and Rybka 3 at 4m2 sec and 25m, gave an almost equal result, 31-19 and 30-20 and are nicely in line with the above statistic findings.
My games with the Nunn2 testmatch (50 games) on long timecontrol 40m/40/20m/40/20
(Centrino Dual Core T8300, 2100Mhz, 128 mb Hash) is yet running at game16, standings sofar, Firebird +6 - 2 =7 so pretty consistent result for Firebird. But we wil see for the complete results within two days whether my assumption - Firebird holds his superiority at longer timecontrols - stays put.
grusse Bram