5.3Single-cycle vs multi-cycle
The idea
A single-cycle CPU does a whole instruction in one long clock; multi-cycle splits it into short cycles; pipelined overlaps instructions.
Think of it like this: One chef cooks each dish start to finish (single), a chef with stations but only one dish at a time (multi), a kitchen line (pipelined).
Why it exists
It explains why CPI and clock speed trade off and why pipelines win.
How it works
- Single-cycle: CPI = 1 but the clock is as slow as the slowest instruction.
- Multi-cycle: clock = slowest stage; each instruction takes 3–5 cycles; hardware is reused.
- Pipelined: short clock and about 1 instruction per cycle.
Worked example: LW, ADD, BEQ with stage time 200 ps
Single-cycle: 3 × 800 = 2400 ps. Multi-cycle: 12 × 200 = 2400 ps. Pipelined: 7 × 200 = 1400 ps.
Now play with it
- →Click through the three tabs and compare how many columns each uses.
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| LW | clk | clk | clk | clk | ||||||||
| ADD | clk | clk | clk | clk | ||||||||
| BEQ | clk | clk | clk | clk |
One long clock (800 ps = slowest instruction, here 4 columns of 200 ps) per instruction: CPI = 1 but a slow clock.
Check yourself
In a single-cycle design the clock period is set by:
Exam notes (the dense version)
Single-cycle: CPI = 1, clock = longest instruction path. Multi-cycle: clock = longest stage, CPI 3–5 and varies. Pipeline gets CPI ≈ 1 with short clock.
Previous year questions: ALU, Datapath & Control Unit (3)
PYQs are grouped by topic, so this is the full set for ALU, Datapath & Control Unit; try the ones that match this subtopic. Open all 3 on the topic page.