This work proposes a charge-transfer-based crossbar, where the accumulation is performed by counting the transferred charges into capacitors. Thanks to the time-discrete property of the charge transfer, the entire proposed design is almost fully dynamic excepting an auxiliary amplifier and very insensitive to parasitic wire resistance. In addition, adaptive reference and body-biasing techniques are used to realize a self-adjustable operating range for neural network computations and stable computation behavior over process corners, respectively. A computation and power efficiency of 1024GOP/s and 78TOPS/w is achieved for computation with 1-bit inputs, 1-bit weight, and 4-bit output.