MIPS and FLOPS Metrics
What you'll learn: How MIPS and FLOPS quantify CPU performance by counting operations executed per second.
The Need for Performance Benchmarks
When comparing processors, we need simple numbers that indicate "how fast" a CPU runs. Two common metrics emerged to capture raw computational power: MIPS (Millions of Instructions Per Second) and FLOPS (Floating-Point Operations Per Second). These metrics help engineers, scientists, and consumers understand processor capabilities at a glance.
MIPS: Counting Instructions
MIPS measures how many millions of instructions a processor executes every second. For example, a 500 MIPS processor completes 500 million instructions per second.
Formula:
MIPS = (Clock Rate) / (CPI × 10⁶)
If your CPU runs at 2 GHz with an average CPI of 2.0:
MIPS = 2,000 MHz / (2.0 × 1) = 1,000 MIPS
The catch: MIPS doesn't account for what instructions do. A complex multiplication counts the same as a simple addition, making cross-architecture comparisons tricky.
FLOPS: Measuring Math Power
FLOPS counts floating-point operations per second—additions, multiplications, divisions with decimal numbers. This metric matters greatly in scientific computing, graphics, and AI workloads where floating-point math dominates.
Modern supercomputers measure performance in teraFLOPS (trillions) or petaFLOPS (quadrillions) of operations per second.
Why separate metrics? Integer-heavy workloads (databases, web servers) care about MIPS. Scientific simulations care about FLOPS. Different applications stress different capabilities.
The Analogy
Think of MIPS as counting how many steps a runner takes per minute, while FLOPS measures how many complex dance moves they perform per minute. Both indicate speed, but for different tasks.
Key Takeaway: MIPS and FLOPS provide rough performance indicators by counting millions of instructions or floating-point operations per second, though they don't capture the full complexity of real-world workload performance.