Bitslice

Bit-slice, or bit-slice architecture, is a design approach in which a processor datapath is built from narrow, identical processing elements that operate on individual bits or small groups of bits and are combined to form wider word sizes. Each slice implements a subset of the processor’s logical, arithmetic, and control functions for a given bit position. The slices are connected so that inter-slice signals such as carries or borrows propagate between adjacent slices, enabling the construction of wider arithmetic units from multiple smaller units. Commonly, bit-slice designs use 4-bit ALU slices, though other widths are used.

In practice, a bit-slice CPU uses a collection of identical 4-bit (or other-width) modules that are joined to create, for example, 8-, 16-, or 32-bit datapaths. Control logic and microcode determine how each slice performs operations in parallel, while inter-slice carry or status signals synchronize across the whole word. Notable historical examples include the 4-bit ALU chip 74181 and the Am2901 family, which served as modular ALU slices that could be assembled into wider CPUs by adding more slices.

Bit-slice architectures were prominent in the 1960s through the 1980s, especially before monolithic wide-width ALUs became standard with advances in integrated circuit technology. Today, bit-slice concepts are mostly of historical interest and are rarely used for new designs, having been superseded by single-chip wide datapaths. They occasionally reappear in cryptographic research and FPGA-based prototyping, where modular, reconfigurable slices remain useful for experimentation.