Instructions Per Cycle (IPC) and Throughput
You'll learn how IPC measures pipeline efficiency by flipping CPI, giving you a direct view of processor throughput.
Understanding IPC
In the previous lesson, you learned about Cycles Per Instruction (CPI) — how many clock cycles it takes to complete one instruction on average. Instructions Per Cycle (IPC) is simply the reciprocal:
IPC = 1 / CPI
If your processor has a CPI of 2.0, then IPC = 0.5 (you complete half an instruction per cycle, on average). If CPI = 0.5, then IPC = 2.0 (you complete two instructions per cycle).
Why IPC Matters for Throughput
Throughput measures how many instructions your processor completes over time. IPC directly tells you this rate. A higher IPC means better throughput — more work done per clock cycle.
Think of it like a factory assembly line: CPI tells you "how many minutes per car," while IPC tells you "how many cars per minute." Both describe the same performance, but IPC naturally expresses productivity and throughput.
Pipeline Efficiency Connection
In an ideal 5-stage pipeline with no hazards, you'd complete one instruction every cycle (IPC = 1.0, CPI = 1.0). But pipeline stalls, bubbles, and hazards hurt performance. If data hazards force stalls, your effective CPI increases to 1.5, meaning IPC drops to 0.67.
IPC is your direct measurement of pipeline efficiency. A well-designed pipeline with good hazard mitigation (forwarding, branch prediction) maximizes IPC. Every time you eliminate a stall, IPC rises and throughput improves.
Key Takeaway: IPC is the inverse of CPI and directly measures how many instructions complete per cycle — making it the clearest metric for understanding processor throughput and pipeline efficiency.