Python performance and measurement

Measure slow work, improve the real bottleneck, and confirm the change helped.

Chapter goal: Measure slow work, improve the real bottleneck, and confirm the change helped.

Simple explanation

Performance asks how much time and memory code uses. Measure first, then improve the part that is actually slow.

In Python, this chapter is about reducing unnecessary work without sacrificing correctness or clarity. Start with the idea above. Then connect each symbol to a value or action in the example.

Do not try to remember every symbol. First ask what data the program has, what it does with that data, and what result it creates. Technical words become easier when you connect them to those three questions.

Why this topic is important

Performance affects usability, battery, server cost, and how many users a system can support. In Python, the syntax may look different from other languages, but the thinking skill transfers: name the data, choose the right operation, and make the next step obvious.

When to use it

Optimize after measurements show slow rendering, repeated queries, heavy loops, memory growth, or blocked input.

Example code

import time

cache = {}

def slow_square(n):
    if n in cache:
        return cache[n]
    time.sleep(0.1)  # pretend this is expensive
    result = n * n
    cache[n] = result
    return result

start = time.time()
slow_square(4)
slow_square(4)  # second call reads the cache instead of repeating the work
print(f"Took {time.time() - start:.2f}s")

Line-by-line explanation

• import time — This brings in code supplied by a library so the example can use its types or functions.
• cache = {} — This stores or updates cache. The value on the right is worked out first, then saved under that name.
• def slow_square(n): — This defines a reusable function. Its name describes the job that other code can call.
• if n in cache: — This checks a true-or-false condition. The controlled block runs only when that condition is true.
• return cache[n] — This sends a result back to the code that called the function. Returning is different from printing.
• time.sleep(0.1) # pretend this is expensive — This line supports one focused piece of application logic. Read it together with the block directly around it.
• result = n * n — This stores or updates result. The value on the right is worked out first, then saved under that name.
• cache[n] = result — This stores or updates cache[n]. The value on the right is worked out first, then saved under that name.
• return result — This sends a result back to the code that called the function. Returning is different from printing.
• start = time.time() — This stores or updates start. The value on the right is worked out first, then saved under that name.

What the output means

The measured result depends on the input size; compare measurements before and after a change.

The output is evidence that the program followed the instructions. If your result is different, read from the first line and write down how each value changes. That is debugging, not failure.

Mistake example

import time

cache = None # optimization is attempted without measuring the slow work

def slow_square(n):
    if n in cache:
        return cache[n]
    time.sleep(0.1)  # pretend this is expensive
    result = n * n
    cache[n] = result
    return result

start = time.time()
slow_square(4)
slow_square(4)  # second call reads the cache instead of repeating the work
print(f"Took {time.time() - start:.2f}s")

This version intentionally shows how optimization is attempted without measuring the slow work. The changed assignment stores a missing value, or a required line is removed, so later code cannot complete its job safely.

Fixed version

import time

cache = {}

def slow_square(n):
    if n in cache:
        return cache[n]
    time.sleep(0.1)  # pretend this is expensive
    result = n * n
    cache[n] = result
    return result

start = time.time()
slow_square(4)
slow_square(4)  # second call reads the cache instead of repeating the work
print(f"Took {time.time() - start:.2f}s")

The corrected version restores the real value or required operation. It fixes the chapter-specific problem: optimization is attempted without measuring the slow work.

Common mistakes

• Optimizing code that is not slow.
• Hiding stale data behind caching.
• Measuring only on a powerful development machine.

Warning: Change one part at a time. If you change many lines together, it becomes harder to learn which change caused the result.

Real use cases

• Reduce unnecessary widget rebuilds.
• Index a database query.
• Avoid repeated work inside a game loop.

Practice exercise

1. Identify repeated work.
2. Measure before changing it.
3. Explain the readability trade-off of the optimization.

Tip: If the exercise feels too large, complete only steps 1 to 3. Small working code teaches more than a large unfinished project.

Mini quiz

1. What measurement proves the code is slow?
2. Which work is repeated unnecessarily?
3. What could caching make stale?

How to read AI-generated code

Do not copy AI code first. Read it like a detective. Find the data, follow the changes, and locate the final output. Ask AI to explain a line only after you have made your own guess.

• What data goes in?
• What values are stored?
• What calculation or decision happens?
• What is printed, displayed, saved, or returned?
• What can go wrong?
• Can you rename one value and still explain the code?

Language reading check

Find the program entry point, follow function calls one at a time, and keep track of each value's type. Do not jump into a class or helper until you know who calls it.

Before you move on

• I can explain this topic in my own words.
• I can read the small example without AI.
• I can change the example and predict the new result.
• I can find and fix one simple mistake.
• I can name one real project that uses this idea.

Next topic

Next, learn deployment and release. Before opening it, explain this chapter out loud in under one minute.