hardwired control

Hardwired control is the other way to build a control unit: instead of looking control signals up in a stored microprogram, you bake the control logic directly into fixed gates and wires. Think of a vending machine whose buttons are physically connected to the right dispensers — press B4 and the candy bar drops, because that path is literally soldered, not looked up in a list. Hardwired control 'wires in' each instruction's signal pattern as permanent combinational (and, for multi-cycle, sequential) logic.

Concretely, you start from the control table — the grid of which signal is 0 or 1 for each opcode — and synthesize it into logic. Each control-signal output becomes a boolean function of the opcode bits, realized as a network of gates (often minimized with a Karnaugh map or a PLA) or, for multi-cycle, a finite-state machine whose states emit the right signals each cycle. There is no internal memory of microinstructions to step through; the signals appear as fast as the gates can settle.

The trade-off is the mirror image of microprogramming. Hardwired control is FAST — pure gate delay, no microcode fetch — and is the natural fit for the simple, regular instruction sets of RISC machines, which is a big part of why RISC could clock high. The downside is rigidity: changing or fixing the control means redesigning logic and re-spinning silicon, not editing a stored table, so it is harder to patch and was historically painful for very complex instruction sets. Real high-performance CPUs use hardwired control for the common, simple instructions and reserve microcode for the rare, complex ones — the practical line between 'hardwired' and 'microprogrammed' is a blend, not a wall.