. CEGT gam# +- CCRL gam# +- diff Houdini 3.0 4CPU 3188 3100 13 3320 1644 17 132 Houdini 2.0c 4CPU 3142 2800 13 3270 2465 13 128 Houdini 1.5 4CPU 3124 4698 9 3260 1582 16 136 Critter 1.2 4CPU 3060 2900 10 3204 1331 17 144 Rybka 4.0 4CPU 3057 4600 8 3191 2389 13 134 Stockfish 2.0.1 4CPU 3033 3000 10 3162 1647 14 129 Houdini 2.0c 1CPU 3033 2700 11 3198 1580 15 165 Rybka 3.0 4CPU 3022 6000 7 3163 5128 10 141 Houdini 1.5 1CPU 3018 5450 8 3166 3039 11 148 Komodo 5.0 1CPU 3001 3300 9 3143 2086 11 142 Critter 1.4 1CPU 2978 2400 11 3139 1663 15 161 Hiarcs 14 4CPU 2975 2000 12 3098 1449 15 123 Rybka 4.0 1CPU 2966 7242 6 3099 2253 13 133 Stockfish 2.2.2 1CPU 2963 2400 18 3112 1647 14 149 Naum 4.2 4CPU 2961 4500 8 3094 4492 9 133 Rybka 3.0 x64 1CPU 2915 5540 8 3077 5998 9 162 Rybka 2.3.2a 4CPU 2907 4150 8 3036 3615 10 129 Shredder 12 4CPU 2900 4600 8 3037 7029 8 137 [avg OA on/off] Naum 4.2 1CPU 2824 9040 6 2978 3496 10 154 Hiarcs 14 1CPU 2823 2800 10 2986 927 19 163 Fritz 13 1CPU 2811 2857 9 2980 2633 11 169 Shredder 12 1CPU 2800 10K 5 2954 3161 11 154 (OA On)
CEGT/CCRL 1cpu/4cpu spreads
CEGT/CCRL 1cpu/4cpu spreads
I found a smattering of top engines (64-bit versions) in 40/4 lists of both CEGT and CCRL, and looked at the difference between the 1cpu and 4cpu versions. My conditions were approximately: needed to be in both lists with same version, and at least 2900 CEGT for 4cpu, and 2800 CEGT for 1cpu. I also didn't include every versions of some multi-appearing programs, and curtailed some data when the error bars were large.