Skip to main content

Performance

Compare relative latencies across the memory hierarchy (5 minutes)​

(CPU registers → L1/L2 cache → RAM → disk → network)

Estimate the time and space complexity of a simple algorithm (20 minutes)​

Recall

In Lecture 3, we mentioned that ArrayList should be your default choice over LinkedList, but deferred the explanation. Now we can understand why: it's all about how modern CPUs interact with memory. ArrayLists store elements contiguously, enabling cache-friendly access patterns that make iteration dramatically faster—even though LinkedLists have O(1) insertion at arbitrary positions "in theory."

Understand garbage collection in Java and how to avoid memory leaks (15 minutes)​

  • How the JVM manages memory (heap, stack, GC roots)
  • Mark-and-sweep basics
  • Common causes of memory leaks (static references, listeners, caches)
  • Tools for detecting memory issues

Apply common patterns to improve performance (15 minutes)​

Caching​

Object pooling and reuse​

Rate-limiting and buffering​

Parallelization​

Performance, Sustainability, and Long-Term Costs (10 minutes)​

Performance optimization is often framed as "making things faster." But in the context of software engineering over time, performance connects to broader sustainability concerns.

Environmental costs compound over time​

  • A 10% efficiency improvement in code run billions of times adds up
  • Energy costs of cloud computing are increasingly visible (and billable)
  • "Green software engineering" is emerging as a discipline and may become a regulatory requirement

Performance choices have distributional effects​

  • Optimizing for high-end devices may exclude users with older hardware
  • As your user base grows, who gets left behind?
  • The users you optimize for in year 1 may not be your primary users in year 5. Or worse: they may still be the same users, but with outdated hardware and software, now frustrated with your app and ready to switch to a competitor.