Re: "Automated Discovery of Search Extensions"
Posted: Tue Dec 07, 2010 2:36 am
For the purposes of comparison I looked at the evaluation Rybka 2.3.2a again. There are a few differences from the Fruit evaluation features as used in Rybka 1.0 Beta. There is "pawn endgame" code, "anti-mobility" for pawns that can't move, no bonus for knight mobility, and a version of the "drawish" indicator from the pawn structure. The 32-bit field to combine two 16-bit opening/endgame scores is present also. Other than these, it seems to be the same as previously, with some of the internal PST scaling from Fruit also still extant.In any case, microwine loads the eval function into 0x46adc0, and the function runs to 0x46bfb3.