mikrokod

Mikrokod, known in English as microcode, is a low-level layer of instructions in a computer’s control unit. It translates machine instructions into sequences of simpler operations that directly manipulate the processor’s datapath, registers, and memory. By delegating complex instructions to microcode, a CPU can implement a rich instruction set with relatively simple, flexible hardware.

Microcode is typically stored in a control store inside the CPU, often implemented as read-only memory or writable memory. A microsequencer or control unit fetches microinstructions and issues the corresponding control signals to the processor. Each microinstruction may specify an operation, the next microinstruction address, and conditional branches based on processor state. The collection of microinstructions forms a microprogram that implements a machine instruction or a part of it.

Historically microcode allowed designers to modify behavior without redesigning hardware. Early computers used hardwired control; microcode emerged in the 1950s–60s and became common for implementing CISC instruction sets. In many processors today, complex instructions are broken down into micro-operations; some CPUs use a microcode interpreter, while others translate instructions into micro-ops on the fly. Microcode can be updated after manufacture to fix bugs or enable new features, delivered via firmware or BIOS updates.

Outcomes and debates: microcode offers flexibility and easier maintenance but can introduce performance overhead compared with fully hardwired control. Modern designs may rely on a combination of microcode and pure hardware control, using micro-ops generated by a decoder rather than a traditional control store. The term mikrokod is widely used in Polish to refer to this concept.