A real-time requirement constrains when a result is produced, not only whether the computed value is correct.

Important timing quantities include:

• latency: time from an event to the relevant response;
• period: intended interval between repeated activities;
• deadline: latest acceptable completion time;
• jitter: variation in the timing of repeated events or responses.

A hard real-time deadline is part of correctness: missing it is considered a system failure. A soft real-time deadline can occasionally be missed with degraded quality rather than immediate incorrectness.

Average execution time is not sufficient when a deadline must always be met. Engineering instead needs useful bounds on worst-case interference, execution time and blocking under the assumed operating conditions.

A fast processor can still fail a real-time requirement if scheduling or blocking produces unpredictable latency.

Real-time design is therefore about bounded temporal behavior rather than simply making software execute as quickly as possible.