Design of Control Unit

Computer Organization & Embedded System — Control and Central Processing Units, NEC licence examination syllabus (Nepal Engineering Council).

Designing the Control Unit itself

Building the machine that reads the microprogram.

Now that we know the microinstruction format, we need actual hardware to read control memory and execute what it says. This is the microprogram sequencer — a small circuit whose only job is to determine the address of the next microinstruction, cycle after cycle.

The sequencer is really just the address-sequencing logic (CAR, mapping, branching, SBR) packaged into one functional block, sitting right next to control memory.

Once a microinstruction is fetched, its F1/F2/F3 fields must be turned into actual control signals — this is done using decoders (classically, three 3-input-to-8-output decoders in Mano's design), each decoder lighting up exactly one wire corresponding to the specific microoperation encoded in that field.

Finally, next-address logic combines the incrementer's output, any branch address, and the subroutine return address — using multiplexers — to decide what value actually loads into CAR for the next cycle.

💡 Be ready to draw the full control-unit block diagram: Control Memory → Microinstruction Register → [F1/F2/F3 → decoders] + [CD/BR/AD → sequencer → next CAR]. This is a favorite long-answer diagram question.

The comparison this section is building toward

Microprogrammed control is one of two ways to build a control unit, and the contrast is a standard long-answer question.

🔀 Hardwired versus microprogrammed

HardwiredControl signals come from fixed combinational logic — gates, decoders, a state machine. Fast, because a signal is one gate delay away, but changing the instruction set means redesigning the circuit.

MicroprogrammedControl signals are read from control memory as microinstructions. Slower, since every step costs a memory read, but the instruction set can be changed by rewriting the microprogram.

This is the same trade-off as custom hardware against a general-purpose processor in the embedded chapter: speed and efficiency against flexibility and development cost. Hardwired control is the custom ASIC of control units; microprogramming is the programmable alternative, cheaper to design and to correct.

💡 It also explains the RISC/CISC split covered later in this subject. CISC's complex, variable-length instructions are practical because a microprogram can implement them; RISC's deliberately simple instructions are chosen so that hardwired control becomes feasible, which is where much of its speed advantage originates.

Why a bug in silicon is not the end

Microprogramming has a practical consequence worth knowing: because the control logic is data in memory rather than fixed wiring, an instruction-set bug found after manufacture can sometimes be corrected by updating the microcode instead of replacing the chip.

That is what a processor microcode update distributed by an operating system actually is — the hardware is unchanged and the control program running inside it is replaced.

💡 Horizontal and vertical microinstruction formats sit on the same axis. Horizontal uses one bit per control signal: wide words, no decoding delay, maximum parallelism. Vertical encodes signals into compact fields needing a decoder: narrower memory, slightly slower. Width against speed, exactly as with hardwired against microprogrammed.

Syllabus points

Microprogram sequencer block diagram
Decoding of microoperation fields
Next-address logic

Create a free account to tick topics off, take notes as you read, watch the video lessons and get a day-by-day study plan built around your exam date.