2025-07-21 · Tae-ho Lee
Profiler homework: what good looks like
Optimization without evidence is guesswork. In Frame Budget Bootcamp, students submit before/after captures with hypothesis notes.
A strong submission identifies the hotspot category: CPU script, GPU overdraw, or memory churn. It shows one change at a time, not a bundle of tweaks that cannot be attributed.
We share anonymized examples where a 30% frame time reduction came from UI canvas batching, not heroic shader work. That reframes optimization as discipline, not heroics.
Students who struggle usually skip the baseline capture. Mentors pair them with a lab slot to rebuild the timeline.