A CMOS VLSI chess microprocessor

The Berkeley Chess Microprocessor (BCM) is a 200000 transistor, 1.2 micron CMOS (N-Well) die, 11 mm by 9 mm in area. It can generate three million legal moves/s. The BCM's novel architecture allows for evaluation of chess board positions several ply deep from the current board position. This chip has three evaluation innovations: (1) pins and X-ray attacks can be determined, (2) dynamic evaluation is possible, and (3) an ALU on each square can sum the values of attacking pieces. The evaluation innovations can be used to produce exponential search speedup for chess programs. This approach of mapping the problem's topology into silicon is general enough to be used to produce exponential speedups in tree searches where interactions are considered in parallel. >