The hidden cost of a loop: profiling for watts, not just milliseconds

Green Coding sounds like a philosophy. In day-to-day engineering it is mostly profiling. The work that burns the most CPU cycles is the work that draws the most power, and finding it is a skill any team can build.

Faster usually means greener

When you cut the number of instructions to serve a request, you cut the energy to serve it. That is why a good flame graph is also a sustainability tool: the tall, wide frames are both your latency and your watts.

Where the waste hides

• Chatty database access — the classic N+1
• Serialisation of data nobody reads
• Polling loops that could be events
• Re-computing what a cache could hold

The cheapest, greenest instruction is the one you never execute.

Make it a budget, not a heroic sprint

One-off optimisation drives regress. We encode efficiency as budgets in CI — a request may not exceed a certain CPU-time or payload size — so the gains stick. Efficiency becomes a property of the codebase, not a quarterly campaign.