Course contentsShow
← React
Lesson 2280 of 2,39650. Performance: Optimization TechniquesFree lesson

Measuring Expensive Computations

Profile render times to determine if a computation is actually slow enough to memoize.

Measuring Expensive Computations

You'll learn: How to determine if a calculation is actually slow enough to warrant useMemo optimization.

The Problem with Guessing

Not all computations are expensive. Wrapping everything in useMemo adds overhead—the hook itself has a cost. You need data, not intuition, to decide if memoization is worth it.

How to Measure

The simplest approach: time it manually using console.time():

function ExpensiveList({ items }) {
  console.time('filterAndSort');
  const processed = items
    .filter(item => item.active)
    .sort((a, b) => a.priority - b.priority);
  console.timeEnd('filterAndSort');

  return <ul>{/* render */}</ul>;
}

Run this and check your browser console. You'll see timing like filterAndSort: 0.2ms or filterAndSort: 45ms.

What's Actually "Expensive"?

Rough guidelines:

  • < 1ms: Almost never worth memoizing
  • 1-5ms: Probably fine, profile in context
  • > 5ms: Consider useMemo if it runs frequently

Remember: a 0.5ms computation that runs 100 times per render adds up differently than one that runs once.

Measuring in Real Scenarios

Profile during realistic interactions:

function DataTable({ rawData }) {
  console.time('processData');
  const processed = rawData.map(/* complex transform */);
  console.timeEnd('processData');
  // Check timing when rawData has 10, 100, 1000, 10000 items
}

Test with production-sized datasets. That cute demo data with 5 items won't reveal the real cost.

The React DevTools Approach

Use the Profiler to see if the computation's render time stands out in the flamegraph. If the component itself is fast (< 2ms total render), the computation probably isn't your bottleneck.

Key Takeaway: Measure computation time with console.time() and realistic data before adding useMemo—optimize based on evidence, not assumptions.