프로파일 보고서 읽기

A profile report lists where your program spent its time, function by function, turning raw runs into clear decisions.

Total time includes everything a function calls; self time is just its own work. Comparing the two reveals the true cost center.

A report shows how often each function ran. A cheap call made millions of times can top the list on count alone.

Order the report by self time so the worst offenders rise to the top. The first row is usually where to start optimizing.

The function eating the largest share is your hot spot. Even a small speedup there moves the whole program forward.

A percentage of total runtime is easier to compare than raw seconds. It instantly shows which line dominates the run.

Reports often expose a slow spot you never suspected. Let the numbers correct your intuition rather than confirming a guess.

Some reports flag time lost waiting on data. Knowing whether the cost is compute or memory points you to the right fix.

Save a report, make a change, then profile again. Comparing the two reports proves whether your optimization really worked.

Tiny differences between runs are just noise. Only trust gaps large and repeatable enough to stand out across several runs.

End every read with one clear action: which function to fix next. A good report becomes a focused optimization plan.

Think about which number to trust in a report.

You learned to read a report: compare total and self time, weigh call counts, sort by cost, ignore noise, and turn it into a plan. 📊