64 Bit Computer Chess Program

Sort:
OneBitPerSquare

Does anyone know if anyone has written a chess program using a 64 bit processor that can "see" the entire chess board in one ("every") machine cycle? Might this approach effectively speed up chess engines by several orders of magnitude? (e.g. 1st test: Is a given chessboard square occupied? [This is useful for evaluating possible (legal) moves, for every playing piece - except a knight.] 2nd test: If occupied then by which color? [If by opponent: This is useful for evaluating threats and attacks. If by me: This is useful for evaluating strategies. ] Etc.) (tm.) By this means couldn't a computer chess engine analyze potential/prospective game moves by the millions into the future in seconds? Or ultimately is this all merely useful as a "compression" engine that stores playing board snapshots using less memory?

arabinativ

 

OneBitPerSquare wrote:

 

Does anyone know if anyone has written a chess program using a 64 bit processor that can "see" the entire chess board in one ("every") machine cycle? Might this approach effectively speed up chess engines by several orders of magnitude? (e.g. 1st test: Is a given chessboard square occupied? [This is useful for evaluating possible (legal) moves, for every playing piece - except a knight.] 2nd test: If occupied then by which color? [If by opponent: This is useful for evaluating threats and attacks. If by me: This is useful for evaluating strategies. ] Etc.) (tm.) By this means couldn't a computer chess engine analyze potential/prospective game moves by the millions into the future in seconds? Or ultimately is this all merely useful as a "compression" engine that stores playing board snapshots using less memory?

 

Unfortunately, about the only condition that might be tested using a single bit per square is occupancy (where, say, 1 means a square is occupied and 0 means it is not), and even then, it wouldn't provide any information as to the color of the piece occupying a particular square, nor what type of piece it might be. For this, additional bits are needed. So no, a direct 1-to-1 bitwise mapping of the board would be of very limited usefulness.