eng-performance

Performance and Efficiency Awareness

Intent

• Treat budgets (frame time, network round-trips, server CPU, on-chain gas) as first-class requirements.
• Prevent regressions by profiling early, not after user complaints.

Inputs

1. Existing SLAs/budgets per platform (FPS, TTI, API latency, contract gas caps).
2. Representative workloads, datasets, replays, or load scripts.
3. Tooling: profilers, flamegraphs, shader analyzers, gas estimators, Lighthouse, etc.

Workflow

1. Establish baseline
   • Capture metrics for the current implementation under realistic load/device tiers.
2. Design with budgets
   • Choose data structures, rendering strategies, and contract patterns aligned with constraints.
   • Consider caching, pagination, batching, and compression aggressively.
3. Measure iteratively
   • After each significant change, run targeted benchmarks or simulations.
   • Track results in the PR/issue; highlight deltas and why they are acceptable.
4. Optimize where it matters
   • Focus on user-facing bottlenecks; avoid premature micro-optimizations.
   • For web3, minimize storage writes and external calls; for mobile, monitor battery/thermal impact.

Verification

• Benchmarks or profiling output attached, showing no regression (or justified improvements).
• Budgets documented; alarms/alerts configured if applicable.
• For regressions that cannot be fixed immediately, open a follow-up issue with owner, impact, and mitigation.